├── EDSR ├── common.py └── edsr.py ├── LICENSE ├── README.md ├── adaptive_gridsampler ├── adaptive_gridsampler_cuda.cpp ├── adaptive_gridsampler_kernel.cu ├── adaptive_gridsampler_kernel.cuh ├── gridsampler.py ├── helper_cuda.h ├── helper_string.h └── setup.py ├── figs ├── overview.png └── qualitative.png ├── modules.py ├── run.py └── utils.py /EDSR/common.py: -------------------------------------------------------------------------------- 1 | import math 2 | 3 | import torch 4 | import torch.nn as nn 5 | import torch.nn.functional as F 6 | 7 | 8 | def default_conv(in_channels, out_channels, kernel_size, bias=True): 9 | return nn.Conv2d( 10 | in_channels, out_channels, kernel_size, 11 | padding=(kernel_size // 2), bias=bias) 12 | 13 | 14 | class MeanShift(nn.Conv2d): 15 | def __init__(self, rgb_range, rgb_mean=(0.4488, 0.4371, 0.4040), rgb_std=(1.0, 1.0, 1.0), sign=-1): 16 | super(MeanShift, self).__init__(3, 3, kernel_size=1) 17 | std = torch.Tensor(rgb_std) 18 | self.weight.data = torch.eye(3).view(3, 3, 1, 1) / std.view(3, 1, 1, 1) 19 | self.bias.data = sign * rgb_range * torch.Tensor(rgb_mean) / std 20 | for p in self.parameters(): 21 | p.requires_grad = False 22 | 23 | 24 | class BasicBlock(nn.Sequential): 25 | def __init__( 26 | self, conv, in_channels, out_channels, kernel_size, stride=1, bias=False, 27 | bn=True, act=nn.ReLU(True)): 28 | 29 | m = [conv(in_channels, out_channels, kernel_size, bias=bias)] 30 | if bn: 31 | m.append(nn.BatchNorm2d(out_channels)) 32 | if act is not None: 33 | m.append(act) 34 | 35 | super(BasicBlock, self).__init__(*m) 36 | 37 | 38 | class ResBlock(nn.Module): 39 | def __init__( 40 | self, conv, n_feats, kernel_size, 41 | bias=True, bn=False, act=nn.ReLU(True), res_scale=1): 42 | 43 | super(ResBlock, self).__init__() 44 | m = [] 45 | for i in range(2): 46 | m.append(conv(n_feats, n_feats, kernel_size, bias=bias)) 47 | if bn: 48 | m.append(nn.BatchNorm2d(n_feats)) 49 | if i == 0: 50 | m.append(act) 51 | 52 | self.body = nn.Sequential(*m) 53 | self.res_scale = res_scale 54 | 55 | def forward(self, x): 56 | res = self.body(x).mul(self.res_scale) 57 | res += x 58 | 59 | return res 60 | 61 | 62 | class Upsampler(nn.Sequential): 63 | def __init__(self, conv, scale, n_feats, bn=False, act=False, bias=True): 64 | 65 | m = [] 66 | if (scale & (scale - 1)) == 0: # Is scale = 2^n? 67 | for _ in range(int(math.log(scale, 2))): 68 | m.append(conv(n_feats, 4 * n_feats, 3, bias)) 69 | m.append(nn.PixelShuffle(2)) 70 | if bn: 71 | m.append(nn.BatchNorm2d(n_feats)) 72 | if act == 'relu': 73 | m.append(nn.ReLU(True)) 74 | elif act == 'prelu': 75 | m.append(nn.PReLU(n_feats)) 76 | 77 | elif scale == 3: 78 | m.append(conv(n_feats, 9 * n_feats, 3, bias)) 79 | m.append(nn.PixelShuffle(3)) 80 | if bn: 81 | m.append(nn.BatchNorm2d(n_feats)) 82 | if act == 'relu': 83 | m.append(nn.ReLU(True)) 84 | elif act == 'prelu': 85 | m.append(nn.PReLU(n_feats)) 86 | else: 87 | raise NotImplementedError 88 | 89 | super(Upsampler, self).__init__(*m) 90 | -------------------------------------------------------------------------------- /EDSR/edsr.py: -------------------------------------------------------------------------------- 1 | from EDSR import common 2 | 3 | import torch 4 | import torch.nn as nn 5 | 6 | url = { 7 | 'r16f64x2': 'https://cv.snu.ac.kr/research/EDSR/models/edsr_baseline_x2-1bc95232.pt', 8 | 'r16f64x3': 'https://cv.snu.ac.kr/research/EDSR/models/edsr_baseline_x3-abf2a44e.pt', 9 | 'r16f64x4': 'https://cv.snu.ac.kr/research/EDSR/models/edsr_baseline_x4-6b446fab.pt', 10 | 'r32f256x2': 'https://cv.snu.ac.kr/research/EDSR/models/edsr_x2-0edfb8a3.pt', 11 | 'r32f256x3': 'https://cv.snu.ac.kr/research/EDSR/models/edsr_x3-ea3ef2c6.pt', 12 | 'r32f256x4': 'https://cv.snu.ac.kr/research/EDSR/models/edsr_x4-4f62e9ef.pt' 13 | } 14 | 15 | 16 | class EDSR(nn.Module): 17 | def __init__(self, n_resblocks=16, n_feats=64, scale=4, conv=common.default_conv): 18 | super(EDSR, self).__init__() 19 | 20 | # n_resblocks = 16 * 2 21 | # n_feats = 64 * 4 22 | kernel_size = 3 23 | act = nn.ReLU(True) 24 | self.url = url['r{}f{}x{}'.format(n_resblocks, n_feats, scale)] 25 | self.sub_mean = common.MeanShift(1) 26 | self.add_mean = common.MeanShift(1, sign=1) 27 | 28 | # define head module 29 | m_head = [conv(3, n_feats, kernel_size)] 30 | 31 | # define body module 32 | m_body = [ 33 | common.ResBlock( 34 | conv, n_feats, kernel_size, act=act, res_scale=0.1 35 | ) for _ in range(n_resblocks) 36 | ] 37 | m_body.append(conv(n_feats, n_feats, kernel_size)) 38 | 39 | # define tail module 40 | m_tail = [ 41 | common.Upsampler(conv, scale, n_feats, act=False), 42 | conv(n_feats, 3, kernel_size) 43 | ] 44 | 45 | self.head = nn.Sequential(*m_head) 46 | self.body = nn.Sequential(*m_body) 47 | self.tail = nn.Sequential(*m_tail) 48 | 49 | def forward(self, x): 50 | x = self.sub_mean(x) 51 | x = self.head(x) 52 | 53 | res = self.body(x) 54 | res += x 55 | 56 | x = self.tail(res) 57 | x = self.add_mean(x) 58 | 59 | return x 60 | 61 | def load_state_dict(self, state_dict, strict=True): 62 | own_state = self.state_dict() 63 | for name, param in state_dict.items(): 64 | if name in own_state: 65 | if isinstance(param, nn.Parameter): 66 | param = param.data 67 | try: 68 | own_state[name].copy_(param) 69 | except Exception: 70 | if name.find('tail') == -1: 71 | raise RuntimeError('While copying the parameter named {}, ' 72 | 'whose dimensions in the model are {} and ' 73 | 'whose dimensions in the checkpoint are {}.' 74 | .format(name, own_state[name].size(), param.size())) 75 | elif strict: 76 | if name.find('tail') == -1: 77 | raise KeyError('unexpected key "{}" in state_dict' 78 | .format(name)) 79 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | GNU GENERAL PUBLIC LICENSE 2 | Version 3, 29 June 2007 3 | 4 | Copyright (C) 2007 Free Software Foundation, Inc. 5 | Everyone is permitted to copy and distribute verbatim copies 6 | of this license document, but changing it is not allowed. 7 | 8 | Preamble 9 | 10 | The GNU General Public License is a free, copyleft license for 11 | software and other kinds of works. 12 | 13 | The licenses for most software and other practical works are designed 14 | to take away your freedom to share and change the works. By contrast, 15 | the GNU General Public License is intended to guarantee your freedom to 16 | share and change all versions of a program--to make sure it remains free 17 | software for all its users. We, the Free Software Foundation, use the 18 | GNU General Public License for most of our software; it applies also to 19 | any other work released this way by its authors. You can apply it to 20 | your programs, too. 21 | 22 | When we speak of free software, we are referring to freedom, not 23 | price. Our General Public Licenses are designed to make sure that you 24 | have the freedom to distribute copies of free software (and charge for 25 | them if you wish), that you receive source code or can get it if you 26 | want it, that you can change the software or use pieces of it in new 27 | free programs, and that you know you can do these things. 28 | 29 | To protect your rights, we need to prevent others from denying you 30 | these rights or asking you to surrender the rights. Therefore, you have 31 | certain responsibilities if you distribute copies of the software, or if 32 | you modify it: responsibilities to respect the freedom of others. 33 | 34 | For example, if you distribute copies of such a program, whether 35 | gratis or for a fee, you must pass on to the recipients the same 36 | freedoms that you received. You must make sure that they, too, receive 37 | or can get the source code. And you must show them these terms so they 38 | know their rights. 39 | 40 | Developers that use the GNU GPL protect your rights with two steps: 41 | (1) assert copyright on the software, and (2) offer you this License 42 | giving you legal permission to copy, distribute and/or modify it. 43 | 44 | For the developers' and authors' protection, the GPL clearly explains 45 | that there is no warranty for this free software. For both users' and 46 | authors' sake, the GPL requires that modified versions be marked as 47 | changed, so that their problems will not be attributed erroneously to 48 | authors of previous versions. 49 | 50 | Some devices are designed to deny users access to install or run 51 | modified versions of the software inside them, although the manufacturer 52 | can do so. This is fundamentally incompatible with the aim of 53 | protecting users' freedom to change the software. The systematic 54 | pattern of such abuse occurs in the area of products for individuals to 55 | use, which is precisely where it is most unacceptable. Therefore, we 56 | have designed this version of the GPL to prohibit the practice for those 57 | products. If such problems arise substantially in other domains, we 58 | stand ready to extend this provision to those domains in future versions 59 | of the GPL, as needed to protect the freedom of users. 60 | 61 | Finally, every program is threatened constantly by software patents. 62 | States should not allow patents to restrict development and use of 63 | software on general-purpose computers, but in those that do, we wish to 64 | avoid the special danger that patents applied to a free program could 65 | make it effectively proprietary. To prevent this, the GPL assures that 66 | patents cannot be used to render the program non-free. 67 | 68 | The precise terms and conditions for copying, distribution and 69 | modification follow. 70 | 71 | TERMS AND CONDITIONS 72 | 73 | 0. Definitions. 74 | 75 | "This License" refers to version 3 of the GNU General Public License. 76 | 77 | "Copyright" also means copyright-like laws that apply to other kinds of 78 | works, such as semiconductor masks. 79 | 80 | "The Program" refers to any copyrightable work licensed under this 81 | License. Each licensee is addressed as "you". "Licensees" and 82 | "recipients" may be individuals or organizations. 83 | 84 | To "modify" a work means to copy from or adapt all or part of the work 85 | in a fashion requiring copyright permission, other than the making of an 86 | exact copy. The resulting work is called a "modified version" of the 87 | earlier work or a work "based on" the earlier work. 88 | 89 | A "covered work" means either the unmodified Program or a work based 90 | on the Program. 91 | 92 | To "propagate" a work means to do anything with it that, without 93 | permission, would make you directly or secondarily liable for 94 | infringement under applicable copyright law, except executing it on a 95 | computer or modifying a private copy. Propagation includes copying, 96 | distribution (with or without modification), making available to the 97 | public, and in some countries other activities as well. 98 | 99 | To "convey" a work means any kind of propagation that enables other 100 | parties to make or receive copies. Mere interaction with a user through 101 | a computer network, with no transfer of a copy, is not conveying. 102 | 103 | An interactive user interface displays "Appropriate Legal Notices" 104 | to the extent that it includes a convenient and prominently visible 105 | feature that (1) displays an appropriate copyright notice, and (2) 106 | tells the user that there is no warranty for the work (except to the 107 | extent that warranties are provided), that licensees may convey the 108 | work under this License, and how to view a copy of this License. If 109 | the interface presents a list of user commands or options, such as a 110 | menu, a prominent item in the list meets this criterion. 111 | 112 | 1. Source Code. 113 | 114 | The "source code" for a work means the preferred form of the work 115 | for making modifications to it. "Object code" means any non-source 116 | form of a work. 117 | 118 | A "Standard Interface" means an interface that either is an official 119 | standard defined by a recognized standards body, or, in the case of 120 | interfaces specified for a particular programming language, one that 121 | is widely used among developers working in that language. 122 | 123 | The "System Libraries" of an executable work include anything, other 124 | than the work as a whole, that (a) is included in the normal form of 125 | packaging a Major Component, but which is not part of that Major 126 | Component, and (b) serves only to enable use of the work with that 127 | Major Component, or to implement a Standard Interface for which an 128 | implementation is available to the public in source code form. A 129 | "Major Component", in this context, means a major essential component 130 | (kernel, window system, and so on) of the specific operating system 131 | (if any) on which the executable work runs, or a compiler used to 132 | produce the work, or an object code interpreter used to run it. 133 | 134 | The "Corresponding Source" for a work in object code form means all 135 | the source code needed to generate, install, and (for an executable 136 | work) run the object code and to modify the work, including scripts to 137 | control those activities. However, it does not include the work's 138 | System Libraries, or general-purpose tools or generally available free 139 | programs which are used unmodified in performing those activities but 140 | which are not part of the work. For example, Corresponding Source 141 | includes interface definition files associated with source files for 142 | the work, and the source code for shared libraries and dynamically 143 | linked subprograms that the work is specifically designed to require, 144 | such as by intimate data communication or control flow between those 145 | subprograms and other parts of the work. 146 | 147 | The Corresponding Source need not include anything that users 148 | can regenerate automatically from other parts of the Corresponding 149 | Source. 150 | 151 | The Corresponding Source for a work in source code form is that 152 | same work. 153 | 154 | 2. Basic Permissions. 155 | 156 | All rights granted under this License are granted for the term of 157 | copyright on the Program, and are irrevocable provided the stated 158 | conditions are met. This License explicitly affirms your unlimited 159 | permission to run the unmodified Program. The output from running a 160 | covered work is covered by this License only if the output, given its 161 | content, constitutes a covered work. This License acknowledges your 162 | rights of fair use or other equivalent, as provided by copyright law. 163 | 164 | You may make, run and propagate covered works that you do not 165 | convey, without conditions so long as your license otherwise remains 166 | in force. You may convey covered works to others for the sole purpose 167 | of having them make modifications exclusively for you, or provide you 168 | with facilities for running those works, provided that you comply with 169 | the terms of this License in conveying all material for which you do 170 | not control copyright. Those thus making or running the covered works 171 | for you must do so exclusively on your behalf, under your direction 172 | and control, on terms that prohibit them from making any copies of 173 | your copyrighted material outside their relationship with you. 174 | 175 | Conveying under any other circumstances is permitted solely under 176 | the conditions stated below. Sublicensing is not allowed; section 10 177 | makes it unnecessary. 178 | 179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law. 180 | 181 | No covered work shall be deemed part of an effective technological 182 | measure under any applicable law fulfilling obligations under article 183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or 184 | similar laws prohibiting or restricting circumvention of such 185 | measures. 186 | 187 | When you convey a covered work, you waive any legal power to forbid 188 | circumvention of technological measures to the extent such circumvention 189 | is effected by exercising rights under this License with respect to 190 | the covered work, and you disclaim any intention to limit operation or 191 | modification of the work as a means of enforcing, against the work's 192 | users, your or third parties' legal rights to forbid circumvention of 193 | technological measures. 194 | 195 | 4. Conveying Verbatim Copies. 196 | 197 | You may convey verbatim copies of the Program's source code as you 198 | receive it, in any medium, provided that you conspicuously and 199 | appropriately publish on each copy an appropriate copyright notice; 200 | keep intact all notices stating that this License and any 201 | non-permissive terms added in accord with section 7 apply to the code; 202 | keep intact all notices of the absence of any warranty; and give all 203 | recipients a copy of this License along with the Program. 204 | 205 | You may charge any price or no price for each copy that you convey, 206 | and you may offer support or warranty protection for a fee. 207 | 208 | 5. Conveying Modified Source Versions. 209 | 210 | You may convey a work based on the Program, or the modifications to 211 | produce it from the Program, in the form of source code under the 212 | terms of section 4, provided that you also meet all of these conditions: 213 | 214 | a) The work must carry prominent notices stating that you modified 215 | it, and giving a relevant date. 216 | 217 | b) The work must carry prominent notices stating that it is 218 | released under this License and any conditions added under section 219 | 7. This requirement modifies the requirement in section 4 to 220 | "keep intact all notices". 221 | 222 | c) You must license the entire work, as a whole, under this 223 | License to anyone who comes into possession of a copy. This 224 | License will therefore apply, along with any applicable section 7 225 | additional terms, to the whole of the work, and all its parts, 226 | regardless of how they are packaged. This License gives no 227 | permission to license the work in any other way, but it does not 228 | invalidate such permission if you have separately received it. 229 | 230 | d) If the work has interactive user interfaces, each must display 231 | Appropriate Legal Notices; however, if the Program has interactive 232 | interfaces that do not display Appropriate Legal Notices, your 233 | work need not make them do so. 234 | 235 | A compilation of a covered work with other separate and independent 236 | works, which are not by their nature extensions of the covered work, 237 | and which are not combined with it such as to form a larger program, 238 | in or on a volume of a storage or distribution medium, is called an 239 | "aggregate" if the compilation and its resulting copyright are not 240 | used to limit the access or legal rights of the compilation's users 241 | beyond what the individual works permit. Inclusion of a covered work 242 | in an aggregate does not cause this License to apply to the other 243 | parts of the aggregate. 244 | 245 | 6. Conveying Non-Source Forms. 246 | 247 | You may convey a covered work in object code form under the terms 248 | of sections 4 and 5, provided that you also convey the 249 | machine-readable Corresponding Source under the terms of this License, 250 | in one of these ways: 251 | 252 | a) Convey the object code in, or embodied in, a physical product 253 | (including a physical distribution medium), accompanied by the 254 | Corresponding Source fixed on a durable physical medium 255 | customarily used for software interchange. 256 | 257 | b) Convey the object code in, or embodied in, a physical product 258 | (including a physical distribution medium), accompanied by a 259 | written offer, valid for at least three years and valid for as 260 | long as you offer spare parts or customer support for that product 261 | model, to give anyone who possesses the object code either (1) a 262 | copy of the Corresponding Source for all the software in the 263 | product that is covered by this License, on a durable physical 264 | medium customarily used for software interchange, for a price no 265 | more than your reasonable cost of physically performing this 266 | conveying of source, or (2) access to copy the 267 | Corresponding Source from a network server at no charge. 268 | 269 | c) Convey individual copies of the object code with a copy of the 270 | written offer to provide the Corresponding Source. This 271 | alternative is allowed only occasionally and noncommercially, and 272 | only if you received the object code with such an offer, in accord 273 | with subsection 6b. 274 | 275 | d) Convey the object code by offering access from a designated 276 | place (gratis or for a charge), and offer equivalent access to the 277 | Corresponding Source in the same way through the same place at no 278 | further charge. You need not require recipients to copy the 279 | Corresponding Source along with the object code. If the place to 280 | copy the object code is a network server, the Corresponding Source 281 | may be on a different server (operated by you or a third party) 282 | that supports equivalent copying facilities, provided you maintain 283 | clear directions next to the object code saying where to find the 284 | Corresponding Source. Regardless of what server hosts the 285 | Corresponding Source, you remain obligated to ensure that it is 286 | available for as long as needed to satisfy these requirements. 287 | 288 | e) Convey the object code using peer-to-peer transmission, provided 289 | you inform other peers where the object code and Corresponding 290 | Source of the work are being offered to the general public at no 291 | charge under subsection 6d. 292 | 293 | A separable portion of the object code, whose source code is excluded 294 | from the Corresponding Source as a System Library, need not be 295 | included in conveying the object code work. 296 | 297 | A "User Product" is either (1) a "consumer product", which means any 298 | tangible personal property which is normally used for personal, family, 299 | or household purposes, or (2) anything designed or sold for incorporation 300 | into a dwelling. In determining whether a product is a consumer product, 301 | doubtful cases shall be resolved in favor of coverage. For a particular 302 | product received by a particular user, "normally used" refers to a 303 | typical or common use of that class of product, regardless of the status 304 | of the particular user or of the way in which the particular user 305 | actually uses, or expects or is expected to use, the product. A product 306 | is a consumer product regardless of whether the product has substantial 307 | commercial, industrial or non-consumer uses, unless such uses represent 308 | the only significant mode of use of the product. 309 | 310 | "Installation Information" for a User Product means any methods, 311 | procedures, authorization keys, or other information required to install 312 | and execute modified versions of a covered work in that User Product from 313 | a modified version of its Corresponding Source. The information must 314 | suffice to ensure that the continued functioning of the modified object 315 | code is in no case prevented or interfered with solely because 316 | modification has been made. 317 | 318 | If you convey an object code work under this section in, or with, or 319 | specifically for use in, a User Product, and the conveying occurs as 320 | part of a transaction in which the right of possession and use of the 321 | User Product is transferred to the recipient in perpetuity or for a 322 | fixed term (regardless of how the transaction is characterized), the 323 | Corresponding Source conveyed under this section must be accompanied 324 | by the Installation Information. But this requirement does not apply 325 | if neither you nor any third party retains the ability to install 326 | modified object code on the User Product (for example, the work has 327 | been installed in ROM). 328 | 329 | The requirement to provide Installation Information does not include a 330 | requirement to continue to provide support service, warranty, or updates 331 | for a work that has been modified or installed by the recipient, or for 332 | the User Product in which it has been modified or installed. Access to a 333 | network may be denied when the modification itself materially and 334 | adversely affects the operation of the network or violates the rules and 335 | protocols for communication across the network. 336 | 337 | Corresponding Source conveyed, and Installation Information provided, 338 | in accord with this section must be in a format that is publicly 339 | documented (and with an implementation available to the public in 340 | source code form), and must require no special password or key for 341 | unpacking, reading or copying. 342 | 343 | 7. Additional Terms. 344 | 345 | "Additional permissions" are terms that supplement the terms of this 346 | License by making exceptions from one or more of its conditions. 347 | Additional permissions that are applicable to the entire Program shall 348 | be treated as though they were included in this License, to the extent 349 | that they are valid under applicable law. If additional permissions 350 | apply only to part of the Program, that part may be used separately 351 | under those permissions, but the entire Program remains governed by 352 | this License without regard to the additional permissions. 353 | 354 | When you convey a copy of a covered work, you may at your option 355 | remove any additional permissions from that copy, or from any part of 356 | it. (Additional permissions may be written to require their own 357 | removal in certain cases when you modify the work.) You may place 358 | additional permissions on material, added by you to a covered work, 359 | for which you have or can give appropriate copyright permission. 360 | 361 | Notwithstanding any other provision of this License, for material you 362 | add to a covered work, you may (if authorized by the copyright holders of 363 | that material) supplement the terms of this License with terms: 364 | 365 | a) Disclaiming warranty or limiting liability differently from the 366 | terms of sections 15 and 16 of this License; or 367 | 368 | b) Requiring preservation of specified reasonable legal notices or 369 | author attributions in that material or in the Appropriate Legal 370 | Notices displayed by works containing it; or 371 | 372 | c) Prohibiting misrepresentation of the origin of that material, or 373 | requiring that modified versions of such material be marked in 374 | reasonable ways as different from the original version; or 375 | 376 | d) Limiting the use for publicity purposes of names of licensors or 377 | authors of the material; or 378 | 379 | e) Declining to grant rights under trademark law for use of some 380 | trade names, trademarks, or service marks; or 381 | 382 | f) Requiring indemnification of licensors and authors of that 383 | material by anyone who conveys the material (or modified versions of 384 | it) with contractual assumptions of liability to the recipient, for 385 | any liability that these contractual assumptions directly impose on 386 | those licensors and authors. 387 | 388 | All other non-permissive additional terms are considered "further 389 | restrictions" within the meaning of section 10. If the Program as you 390 | received it, or any part of it, contains a notice stating that it is 391 | governed by this License along with a term that is a further 392 | restriction, you may remove that term. If a license document contains 393 | a further restriction but permits relicensing or conveying under this 394 | License, you may add to a covered work material governed by the terms 395 | of that license document, provided that the further restriction does 396 | not survive such relicensing or conveying. 397 | 398 | If you add terms to a covered work in accord with this section, you 399 | must place, in the relevant source files, a statement of the 400 | additional terms that apply to those files, or a notice indicating 401 | where to find the applicable terms. 402 | 403 | Additional terms, permissive or non-permissive, may be stated in the 404 | form of a separately written license, or stated as exceptions; 405 | the above requirements apply either way. 406 | 407 | 8. Termination. 408 | 409 | You may not propagate or modify a covered work except as expressly 410 | provided under this License. Any attempt otherwise to propagate or 411 | modify it is void, and will automatically terminate your rights under 412 | this License (including any patent licenses granted under the third 413 | paragraph of section 11). 414 | 415 | However, if you cease all violation of this License, then your 416 | license from a particular copyright holder is reinstated (a) 417 | provisionally, unless and until the copyright holder explicitly and 418 | finally terminates your license, and (b) permanently, if the copyright 419 | holder fails to notify you of the violation by some reasonable means 420 | prior to 60 days after the cessation. 421 | 422 | Moreover, your license from a particular copyright holder is 423 | reinstated permanently if the copyright holder notifies you of the 424 | violation by some reasonable means, this is the first time you have 425 | received notice of violation of this License (for any work) from that 426 | copyright holder, and you cure the violation prior to 30 days after 427 | your receipt of the notice. 428 | 429 | Termination of your rights under this section does not terminate the 430 | licenses of parties who have received copies or rights from you under 431 | this License. If your rights have been terminated and not permanently 432 | reinstated, you do not qualify to receive new licenses for the same 433 | material under section 10. 434 | 435 | 9. Acceptance Not Required for Having Copies. 436 | 437 | You are not required to accept this License in order to receive or 438 | run a copy of the Program. Ancillary propagation of a covered work 439 | occurring solely as a consequence of using peer-to-peer transmission 440 | to receive a copy likewise does not require acceptance. However, 441 | nothing other than this License grants you permission to propagate or 442 | modify any covered work. These actions infringe copyright if you do 443 | not accept this License. Therefore, by modifying or propagating a 444 | covered work, you indicate your acceptance of this License to do so. 445 | 446 | 10. Automatic Licensing of Downstream Recipients. 447 | 448 | Each time you convey a covered work, the recipient automatically 449 | receives a license from the original licensors, to run, modify and 450 | propagate that work, subject to this License. You are not responsible 451 | for enforcing compliance by third parties with this License. 452 | 453 | An "entity transaction" is a transaction transferring control of an 454 | organization, or substantially all assets of one, or subdividing an 455 | organization, or merging organizations. If propagation of a covered 456 | work results from an entity transaction, each party to that 457 | transaction who receives a copy of the work also receives whatever 458 | licenses to the work the party's predecessor in interest had or could 459 | give under the previous paragraph, plus a right to possession of the 460 | Corresponding Source of the work from the predecessor in interest, if 461 | the predecessor has it or can get it with reasonable efforts. 462 | 463 | You may not impose any further restrictions on the exercise of the 464 | rights granted or affirmed under this License. For example, you may 465 | not impose a license fee, royalty, or other charge for exercise of 466 | rights granted under this License, and you may not initiate litigation 467 | (including a cross-claim or counterclaim in a lawsuit) alleging that 468 | any patent claim is infringed by making, using, selling, offering for 469 | sale, or importing the Program or any portion of it. 470 | 471 | 11. Patents. 472 | 473 | A "contributor" is a copyright holder who authorizes use under this 474 | License of the Program or a work on which the Program is based. The 475 | work thus licensed is called the contributor's "contributor version". 476 | 477 | A contributor's "essential patent claims" are all patent claims 478 | owned or controlled by the contributor, whether already acquired or 479 | hereafter acquired, that would be infringed by some manner, permitted 480 | by this License, of making, using, or selling its contributor version, 481 | but do not include claims that would be infringed only as a 482 | consequence of further modification of the contributor version. For 483 | purposes of this definition, "control" includes the right to grant 484 | patent sublicenses in a manner consistent with the requirements of 485 | this License. 486 | 487 | Each contributor grants you a non-exclusive, worldwide, royalty-free 488 | patent license under the contributor's essential patent claims, to 489 | make, use, sell, offer for sale, import and otherwise run, modify and 490 | propagate the contents of its contributor version. 491 | 492 | In the following three paragraphs, a "patent license" is any express 493 | agreement or commitment, however denominated, not to enforce a patent 494 | (such as an express permission to practice a patent or covenant not to 495 | sue for patent infringement). To "grant" such a patent license to a 496 | party means to make such an agreement or commitment not to enforce a 497 | patent against the party. 498 | 499 | If you convey a covered work, knowingly relying on a patent license, 500 | and the Corresponding Source of the work is not available for anyone 501 | to copy, free of charge and under the terms of this License, through a 502 | publicly available network server or other readily accessible means, 503 | then you must either (1) cause the Corresponding Source to be so 504 | available, or (2) arrange to deprive yourself of the benefit of the 505 | patent license for this particular work, or (3) arrange, in a manner 506 | consistent with the requirements of this License, to extend the patent 507 | license to downstream recipients. "Knowingly relying" means you have 508 | actual knowledge that, but for the patent license, your conveying the 509 | covered work in a country, or your recipient's use of the covered work 510 | in a country, would infringe one or more identifiable patents in that 511 | country that you have reason to believe are valid. 512 | 513 | If, pursuant to or in connection with a single transaction or 514 | arrangement, you convey, or propagate by procuring conveyance of, a 515 | covered work, and grant a patent license to some of the parties 516 | receiving the covered work authorizing them to use, propagate, modify 517 | or convey a specific copy of the covered work, then the patent license 518 | you grant is automatically extended to all recipients of the covered 519 | work and works based on it. 520 | 521 | A patent license is "discriminatory" if it does not include within 522 | the scope of its coverage, prohibits the exercise of, or is 523 | conditioned on the non-exercise of one or more of the rights that are 524 | specifically granted under this License. You may not convey a covered 525 | work if you are a party to an arrangement with a third party that is 526 | in the business of distributing software, under which you make payment 527 | to the third party based on the extent of your activity of conveying 528 | the work, and under which the third party grants, to any of the 529 | parties who would receive the covered work from you, a discriminatory 530 | patent license (a) in connection with copies of the covered work 531 | conveyed by you (or copies made from those copies), or (b) primarily 532 | for and in connection with specific products or compilations that 533 | contain the covered work, unless you entered into that arrangement, 534 | or that patent license was granted, prior to 28 March 2007. 535 | 536 | Nothing in this License shall be construed as excluding or limiting 537 | any implied license or other defenses to infringement that may 538 | otherwise be available to you under applicable patent law. 539 | 540 | 12. No Surrender of Others' Freedom. 541 | 542 | If conditions are imposed on you (whether by court order, agreement or 543 | otherwise) that contradict the conditions of this License, they do not 544 | excuse you from the conditions of this License. If you cannot convey a 545 | covered work so as to satisfy simultaneously your obligations under this 546 | License and any other pertinent obligations, then as a consequence you may 547 | not convey it at all. For example, if you agree to terms that obligate you 548 | to collect a royalty for further conveying from those to whom you convey 549 | the Program, the only way you could satisfy both those terms and this 550 | License would be to refrain entirely from conveying the Program. 551 | 552 | 13. Use with the GNU Affero General Public License. 553 | 554 | Notwithstanding any other provision of this License, you have 555 | permission to link or combine any covered work with a work licensed 556 | under version 3 of the GNU Affero General Public License into a single 557 | combined work, and to convey the resulting work. The terms of this 558 | License will continue to apply to the part which is the covered work, 559 | but the special requirements of the GNU Affero General Public License, 560 | section 13, concerning interaction through a network will apply to the 561 | combination as such. 562 | 563 | 14. Revised Versions of this License. 564 | 565 | The Free Software Foundation may publish revised and/or new versions of 566 | the GNU General Public License from time to time. Such new versions will 567 | be similar in spirit to the present version, but may differ in detail to 568 | address new problems or concerns. 569 | 570 | Each version is given a distinguishing version number. If the 571 | Program specifies that a certain numbered version of the GNU General 572 | Public License "or any later version" applies to it, you have the 573 | option of following the terms and conditions either of that numbered 574 | version or of any later version published by the Free Software 575 | Foundation. If the Program does not specify a version number of the 576 | GNU General Public License, you may choose any version ever published 577 | by the Free Software Foundation. 578 | 579 | If the Program specifies that a proxy can decide which future 580 | versions of the GNU General Public License can be used, that proxy's 581 | public statement of acceptance of a version permanently authorizes you 582 | to choose that version for the Program. 583 | 584 | Later license versions may give you additional or different 585 | permissions. However, no additional obligations are imposed on any 586 | author or copyright holder as a result of your choosing to follow a 587 | later version. 588 | 589 | 15. Disclaimer of Warranty. 590 | 591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY 592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT 593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY 594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, 595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM 597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF 598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 599 | 600 | 16. Limitation of Liability. 601 | 602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING 603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS 604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY 605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE 606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF 607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD 608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), 609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF 610 | SUCH DAMAGES. 611 | 612 | 17. Interpretation of Sections 15 and 16. 613 | 614 | If the disclaimer of warranty and limitation of liability provided 615 | above cannot be given local legal effect according to their terms, 616 | reviewing courts shall apply local law that most closely approximates 617 | an absolute waiver of all civil liability in connection with the 618 | Program, unless a warranty or assumption of liability accompanies a 619 | copy of the Program in return for a fee. 620 | 621 | END OF TERMS AND CONDITIONS 622 | 623 | How to Apply These Terms to Your New Programs 624 | 625 | If you develop a new program, and you want it to be of the greatest 626 | possible use to the public, the best way to achieve this is to make it 627 | free software which everyone can redistribute and change under these terms. 628 | 629 | To do so, attach the following notices to the program. It is safest 630 | to attach them to the start of each source file to most effectively 631 | state the exclusion of warranty; and each file should have at least 632 | the "copyright" line and a pointer to where the full notice is found. 633 | 634 | 635 | Copyright (C) 636 | 637 | This program is free software: you can redistribute it and/or modify 638 | it under the terms of the GNU General Public License as published by 639 | the Free Software Foundation, either version 3 of the License, or 640 | (at your option) any later version. 641 | 642 | This program is distributed in the hope that it will be useful, 643 | but WITHOUT ANY WARRANTY; without even the implied warranty of 644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 645 | GNU General Public License for more details. 646 | 647 | You should have received a copy of the GNU General Public License 648 | along with this program. If not, see . 649 | 650 | Also add information on how to contact you by electronic and paper mail. 651 | 652 | If the program does terminal interaction, make it output a short 653 | notice like this when it starts in an interactive mode: 654 | 655 | Copyright (C) 656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'. 657 | This is free software, and you are welcome to redistribute it 658 | under certain conditions; type `show c' for details. 659 | 660 | The hypothetical commands `show w' and `show c' should show the appropriate 661 | parts of the General Public License. Of course, your program's commands 662 | might be different; for a GUI interface, you would use an "about box". 663 | 664 | You should also get your employer (if you work as a programmer) or school, 665 | if any, to sign a "copyright disclaimer" for the program, if necessary. 666 | For more information on this, and how to apply and follow the GNU GPL, see 667 | . 668 | 669 | The GNU General Public License does not permit incorporating your program 670 | into proprietary programs. If your program is a subroutine library, you 671 | may consider it more useful to permit linking proprietary applications with 672 | the library. If this is what you want to do, use the GNU Lesser General 673 | Public License instead of this License. But first, please read 674 | . -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # CAR-pytorch 2 | 3 | Pytorch implementation of paper **"Learned Image Downscaling for Upscaling using Content Adaptive Resampler"** 4 | 5 | ![](figs/overview.png) 6 | 7 | ## Installation 8 | 9 | # get CAR-pytorch source 10 | git clone https://github.com/sunwj/CAR.git 11 | cd CAR 12 | 13 | # compile the code of the resampler 14 | cd adaptive_gridsampler 15 | python3 setup.py build_ext --inplace 16 | 17 | ### Python requirements 18 | Currently, the code only supports python3 and machine with NVIDIA GPU (and the CUDA development toolkit) installed 19 | 20 | * numpy 21 | * scipy 22 | * pytorch (== 1.3.1) 23 | * Pillow 24 | * tqdm 25 | 26 | ### Pre-trained models 27 | You can download the pre-trained models for 2x and 4x downscaling and super-resolution from [here](https://mega.nz/#!XzIm3YhT!jbIOOOGBOiKtv3VAOD782Mz7nK1L_kma-BzR-RhboW4). 28 | 29 | ## Inference 30 | python3 run.py --scale 4 --img_dir path_to_images --model_dir path_to_pretrained_models \ 31 | --output_dir path_to_output 32 | 33 | ## Sample results 34 | ![](figs/qualitative.png) 35 | 36 | You can download HR images of benchmark datasets, i.e., the Set5, Set14, B100 and Urban100 from [here](https://mega.nz/#!znBRCSJA!_qwJMP5VDe3yleiK8m0QXrpHLee9AS8vzT03lAOorP0). 37 | 38 | If you find our work useful in your research or publication, please cite our work: 39 | 40 | Wanjie Sun, Zhenzhong Chen. **"Learned Image Downscaling for Upscaling using Content Adaptive Resampler"**. arXiv preprint arXiv:1907.12904, 2019. 41 | 42 | ``` 43 | @article{sun2020learned, 44 | title={Learned image downscaling for upscaling using content adaptive resampler}, 45 | author={Sun, Wanjie and Chen, Zhenzhong}, 46 | journal={IEEE Transactions on Image Processing}, 47 | volume={29}, 48 | pages={4027--4040}, 49 | year={2020}, 50 | publisher={IEEE} 51 | } 52 | ``` 53 | 54 | ## Acknowlegements 55 | EDSR code is provided by [thstkdgus35/EDSR-PyTorch](https://github.com/thstkdgus35/EDSR-PyTorch). 56 | -------------------------------------------------------------------------------- /adaptive_gridsampler/adaptive_gridsampler_cuda.cpp: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | 4 | #include "adaptive_gridsampler_kernel.cuh" 5 | 6 | int adaptive_gridsampler_cuda_forward(at::Tensor& img, at::Tensor& kernels, at::Tensor& offsets_h, at::Tensor& offsets_v, int offset_unit, int padding, at::Tensor& output) 7 | { 8 | adaptive_gridsampler_kernel_forward(img, kernels, offsets_h, offsets_v, offset_unit, padding, output); 9 | return 1; 10 | } 11 | 12 | int adaptive_gridsampler_cuda_backward(at::Tensor& img, at::Tensor& kernels, at::Tensor& offsets_h, at::Tensor& offsets_v, int offset_unit, at::Tensor& gradOutput, int padding, 13 | at::Tensor& gradInput_kernels, at::Tensor& gradInput_offsets_h, at::Tensor& gradInput_offsets_v) 14 | { 15 | adaptive_gridsampler_kernel_backward(img, kernels, offsets_h, offsets_v, offset_unit, gradOutput, padding, gradInput_kernels, gradInput_offsets_h, gradInput_offsets_v); 16 | return 1; 17 | } 18 | 19 | PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) 20 | { 21 | m.def("forward", &adaptive_gridsampler_cuda_forward, "adaptive gridsampler forward (CUDA)"); 22 | m.def("backward", &adaptive_gridsampler_cuda_backward, "adaptive gridsampler backward (CUDA)"); 23 | } -------------------------------------------------------------------------------- /adaptive_gridsampler/adaptive_gridsampler_kernel.cu: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | 4 | #include "helper_cuda.h" 5 | 6 | #define BLOCK_SIZE 256 7 | 8 | template 9 | __global__ void kernel_adaptive_gridsampler_update_output( 10 | const torch::PackedTensorAccessor32 img, 11 | const torch::PackedTensorAccessor32 kernels, 12 | const torch::PackedTensorAccessor32 offsets_h, 13 | const torch::PackedTensorAccessor32 offsets_v, 14 | const int offset_unit, 15 | const int padding, 16 | torch::PackedTensorAccessor32 output, 17 | const size_t n) 18 | { 19 | auto global_idx = blockDim.x * blockIdx.x + threadIdx.x; 20 | if(global_idx >= n) return; 21 | 22 | auto dim_b = output.size(0); 23 | auto dim_c = output.size(1); 24 | auto dim_h = output.size(2); 25 | auto dim_w = output.size(3); 26 | 27 | auto idb = (global_idx / (dim_c * dim_h * dim_w)) % dim_b; 28 | auto idc = (global_idx / (dim_h * dim_w)) % dim_c; 29 | auto idy = (global_idx / dim_w) % dim_h; 30 | auto idx = global_idx % dim_w; 31 | 32 | if(idx >= dim_w || idy >= dim_h) 33 | return; 34 | 35 | int k_size = sqrt(float(kernels.size(1))); 36 | float w = float(img.size(3) - 2 * padding); 37 | float h = float(img.size(2) - 2 * padding); 38 | 39 | scalar_t result = 0; 40 | for(int k_y = 0; k_y < k_size; ++k_y) 41 | { 42 | for(int k_x = 0; k_x < k_size; ++k_x) 43 | { 44 | scalar_t offset_h = offsets_h[idb][k_size * k_y + k_x][idy][idx] * offset_unit; 45 | scalar_t offset_v = offsets_v[idb][k_size * k_y + k_x][idy][idx] * offset_unit; 46 | 47 | scalar_t p_x = static_cast(idx + 0.5) / dim_w * w + k_x + offset_h - 0.5; 48 | scalar_t p_y = static_cast(idy + 0.5) / dim_h * h + k_y + offset_v - 0.5; 49 | scalar_t alpha = p_x - floor(p_x); 50 | scalar_t beta = p_y - floor(p_y); 51 | 52 | int xL = max(min(int(floor(p_x)), int(w + 2 * padding - 1)), 0); 53 | int xR = max(min(xL + 1, int(w + 2 * padding - 1)), 0); 54 | int yT = max(min(int(floor(p_y)), int(h + 2 * padding - 1)), 0); 55 | int yB = max(min(yT + 1, int(h + 2 * padding - 1)), 0); 56 | 57 | scalar_t val = 0; 58 | val += (1 - alpha) * (1 - beta) * img[idb][idc][yT][xL]; 59 | val += alpha * (1 - beta) * img[idb][idc][yT][xR]; 60 | val += (1 - alpha) * beta * img[idb][idc][yB][xL]; 61 | val += alpha * beta * img[idb][idc][yB][xR]; 62 | 63 | result += val * kernels[idb][k_size * k_y + k_x][idy][idx]; 64 | } 65 | } 66 | output[idb][idc][idy][idx] = result; 67 | } 68 | 69 | void adaptive_gridsampler_kernel_forward(const torch::Tensor& img, const torch::Tensor& kernels, const torch::Tensor& offsets_h, const torch::Tensor& offsets_v, const int offset_unit, const int padding, torch::Tensor& output) 70 | { 71 | kernel_adaptive_gridsampler_update_output<<<(output.numel() + BLOCK_SIZE - 1) / BLOCK_SIZE, BLOCK_SIZE>>>( 72 | img.packed_accessor32(), kernels.packed_accessor32(), 73 | offsets_h.packed_accessor32(), offsets_v.packed_accessor32(), offset_unit, padding, 74 | output.packed_accessor32(), output.numel()); 75 | 76 | checkCudaErrors(cudaGetLastError()); 77 | } 78 | 79 | template 80 | __global__ void kernel_adaptive_gridsampler_backward(const torch::PackedTensorAccessor32 img, 81 | const torch::PackedTensorAccessor32 kernels, 82 | const torch::PackedTensorAccessor32 offsets_h, 83 | const torch::PackedTensorAccessor32 offsets_v, 84 | const int offset_unit, 85 | const torch::PackedTensorAccessor32 gradOutput, 86 | const int padding, 87 | torch::PackedTensorAccessor32 gradInput_kernels, 88 | torch::PackedTensorAccessor32 gradInput_offsets_h, 89 | torch::PackedTensorAccessor32 gradInput_offsets_v, 90 | const size_t n) 91 | { 92 | auto global_idx = blockDim.x * blockIdx.x + threadIdx.x; 93 | if(global_idx >= n) return; 94 | 95 | auto dim_b = gradInput_kernels.size(0); 96 | auto dim_c = gradInput_kernels.size(1); 97 | auto dim_h = gradInput_kernels.size(2); 98 | auto dim_w = gradInput_kernels.size(3); 99 | 100 | auto idb = (global_idx / (dim_c * dim_h * dim_w)) % dim_b; 101 | auto idc = (global_idx / (dim_h * dim_w)) % dim_c; 102 | auto idy = (global_idx / dim_w) % dim_h; 103 | auto idx = global_idx % dim_w; 104 | 105 | if(idx >= dim_w || idx >= dim_h) 106 | return; 107 | 108 | int k_size = sqrt(float(dim_c)); 109 | int k_y = idc / k_size; 110 | int k_x = idc % k_size; 111 | 112 | scalar_t offset_h = offsets_h[idb][idc][idy][idx] * offset_unit; 113 | scalar_t offset_v = offsets_v[idb][idc][idy][idx] * offset_unit; 114 | 115 | float w = float(img.size(3) - 2 * padding); 116 | float h = float(img.size(2) - 2 * padding); 117 | 118 | scalar_t p_x = static_cast(idx + 0.5) / dim_w * w + k_x + offset_h - 0.5; 119 | scalar_t p_y = static_cast(idy + 0.5) / dim_h * h + k_y + offset_v - 0.5; 120 | scalar_t alpha = p_x - floor(p_x); 121 | scalar_t beta = p_y - floor(p_y); 122 | 123 | int xL = max(min(int(floor(p_x)), int(w + 2 * padding - 1)), 0); 124 | int xR = max(min(xL + 1, int(w + 2 * padding - 1)), 0); 125 | int yT = max(min(int(floor(p_y)), int(h + 2 * padding - 1)), 0); 126 | int yB = max(min(yT + 1, int(h + 2 * padding - 1)), 0); 127 | 128 | scalar_t grad_kernels = 0; 129 | scalar_t grad_offset_h = 0; 130 | scalar_t grad_offset_v = 0; 131 | for(int c = 0; c < img.size(1); ++c) 132 | { 133 | scalar_t c_tl = img[idb][c][yT][xL]; 134 | scalar_t c_tr = img[idb][c][yT][xR]; 135 | scalar_t c_bl = img[idb][c][yB][xL]; 136 | scalar_t c_br = img[idb][c][yB][xR]; 137 | 138 | scalar_t grad = 0; 139 | grad += (1 - alpha) * (1 - beta) * c_tl; 140 | grad += alpha * (1 - beta) * c_tr; 141 | grad += (1 - alpha) * beta * c_bl; 142 | grad += alpha * beta * c_br; 143 | grad_kernels += grad * gradOutput[idb][c][idy][idx]; 144 | 145 | grad = (beta - 1) * c_tl + (1 - beta) * c_tr - beta * c_bl + beta * c_br; 146 | grad_offset_h += kernels[idb][idc][idy][idx] * grad * gradOutput[idb][c][idy][idx] * offset_unit; 147 | 148 | grad = (alpha - 1) * c_tl - alpha * c_tr + (1 - alpha) * c_bl + alpha * c_br; 149 | grad_offset_v += kernels[idb][idc][idy][idx] * grad * gradOutput[idb][c][idy][idx] * offset_unit; 150 | } 151 | 152 | gradInput_kernels[idb][idc][idy][idx] = grad_kernels; 153 | 154 | gradInput_offsets_h[idb][idc][idy][idx] = grad_offset_h; 155 | gradInput_offsets_v[idb][idc][idy][idx] = grad_offset_v; 156 | } 157 | 158 | void adaptive_gridsampler_kernel_backward(const torch::Tensor& img, const torch::Tensor& kernels, const torch::Tensor& offsets_h, const torch::Tensor& offsets_v, const int offset_unit, const torch::Tensor& gradOutput, const int padding, 159 | torch::Tensor& gradInput_kernels, torch::Tensor& gradInput_offsets_h, torch::Tensor& gradInput_offsets_v) 160 | { 161 | kernel_adaptive_gridsampler_backward<<<(gradInput_kernels.numel() + BLOCK_SIZE - 1) / BLOCK_SIZE, BLOCK_SIZE, 0>>>( 162 | img.packed_accessor32(), kernels.packed_accessor32(), 163 | offsets_h.packed_accessor32(), offsets_v.packed_accessor32(), 164 | offset_unit, 165 | gradOutput.packed_accessor32(), 166 | padding, 167 | gradInput_kernels.packed_accessor32(), 168 | gradInput_offsets_h.packed_accessor32(), gradInput_offsets_v.packed_accessor32(), 169 | gradInput_kernels.numel()); 170 | 171 | checkCudaErrors(cudaGetLastError()); 172 | } -------------------------------------------------------------------------------- /adaptive_gridsampler/adaptive_gridsampler_kernel.cuh: -------------------------------------------------------------------------------- 1 | #ifndef ADAPTIVE_GRIDSAMPLER_KERNEL_CUH 2 | #define ADAPTIVE_GRIDSAMPLER_KERNEL_CUH 3 | 4 | #include 5 | 6 | void adaptive_gridsampler_kernel_forward(const torch::Tensor& img, const torch::Tensor& kernels, const torch::Tensor& offsets_h, const torch::Tensor& offsets_v, const int offset_unit, const int padding, torch::Tensor& output); 7 | void adaptive_gridsampler_kernel_backward(const torch::Tensor& img, const torch::Tensor& kernels, const torch::Tensor& offsets_h, const torch::Tensor& offsets_v, const int offset_unit, const torch::Tensor& gradOutput, const int padding, torch::Tensor& gradInput_kernels, torch::Tensor& gradInput_offsets_h, torch::Tensor& gradInput_offsets_v); 8 | 9 | #endif -------------------------------------------------------------------------------- /adaptive_gridsampler/gridsampler.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | import torch.nn.functional as F 4 | from torch.autograd import Function, gradcheck 5 | 6 | from .adaptive_gridsampler_cuda import forward 7 | 8 | 9 | class GridSamplerFunction(Function): 10 | @staticmethod 11 | def forward(ctx, img, kernels, offsets_h, offsets_v, offset_unit, padding, downscale_factor): 12 | assert isinstance(downscale_factor, int) 13 | assert isinstance(padding, int) 14 | 15 | ctx.padding = padding 16 | ctx.offset_unit = offset_unit 17 | 18 | b, c, h, w = img.size() 19 | assert h // downscale_factor == kernels.size(2) 20 | assert w // downscale_factor == kernels.size(3) 21 | 22 | img = nn.ReflectionPad2d(padding)(img) 23 | # ctx.save_for_backward(img, kernels, offsets_h, offsets_v) 24 | 25 | output = img.new(b, c, h // downscale_factor, w // downscale_factor).zero_() 26 | forward(img, kernels, offsets_h, offsets_v, offset_unit, padding, output) 27 | 28 | return output 29 | 30 | @staticmethod 31 | def backward(ctx, grad_output): 32 | raise NotImplementedError 33 | 34 | 35 | class Downsampler(nn.Module): 36 | def __init__(self, ds, k_size): 37 | super(Downsampler, self).__init__() 38 | self.ds = ds 39 | self.k_size = k_size 40 | 41 | def forward(self, img, kernels, offsets_h, offsets_v, offset_unit): 42 | assert self.k_size ** 2 == kernels.size(1) 43 | return GridSamplerFunction.apply(img, kernels, offsets_h, offsets_v, offset_unit, self.k_size // 2, self.ds) 44 | -------------------------------------------------------------------------------- /adaptive_gridsampler/helper_cuda.h: -------------------------------------------------------------------------------- 1 | /** 2 | * Copyright 1993-2012 NVIDIA Corporation. All rights reserved. 3 | * 4 | * Please refer to the NVIDIA end user license agreement (EULA) associated 5 | * with this source code for terms and conditions that govern your use of 6 | * this software. Any use, reproduction, disclosure, or distribution of 7 | * this software and related documentation outside the terms of the EULA 8 | * is strictly prohibited. 9 | * 10 | */ 11 | 12 | //////////////////////////////////////////////////////////////////////////////// 13 | // These are CUDA Helper functions for initialization and error checking 14 | 15 | #ifndef HELPER_CUDA_H 16 | #define HELPER_CUDA_H 17 | 18 | #pragma once 19 | 20 | #include 21 | #include 22 | #include 23 | 24 | #include "helper_string.h" 25 | 26 | //#include 27 | //#include 28 | //#include 29 | 30 | // Note, it is required that your SDK sample to include the proper header files, please 31 | // refer the CUDA examples for examples of the needed CUDA headers, which may change depending 32 | // on which CUDA functions are used. 33 | 34 | // CUDA Runtime error messages 35 | #ifdef __DRIVER_TYPES_H__ 36 | static const char *_cudaGetErrorEnum(cudaError_t error) 37 | { 38 | switch (error) 39 | { 40 | case cudaSuccess: 41 | return "cudaSuccess"; 42 | 43 | case cudaErrorMissingConfiguration: 44 | return "cudaErrorMissingConfiguration"; 45 | 46 | case cudaErrorMemoryAllocation: 47 | return "cudaErrorMemoryAllocation"; 48 | 49 | case cudaErrorInitializationError: 50 | return "cudaErrorInitializationError"; 51 | 52 | case cudaErrorLaunchFailure: 53 | return "cudaErrorLaunchFailure"; 54 | 55 | case cudaErrorPriorLaunchFailure: 56 | return "cudaErrorPriorLaunchFailure"; 57 | 58 | case cudaErrorLaunchTimeout: 59 | return "cudaErrorLaunchTimeout"; 60 | 61 | case cudaErrorLaunchOutOfResources: 62 | return "cudaErrorLaunchOutOfResources"; 63 | 64 | case cudaErrorInvalidDeviceFunction: 65 | return "cudaErrorInvalidDeviceFunction"; 66 | 67 | case cudaErrorInvalidConfiguration: 68 | return "cudaErrorInvalidConfiguration"; 69 | 70 | case cudaErrorInvalidDevice: 71 | return "cudaErrorInvalidDevice"; 72 | 73 | case cudaErrorInvalidValue: 74 | return "cudaErrorInvalidValue"; 75 | 76 | case cudaErrorInvalidPitchValue: 77 | return "cudaErrorInvalidPitchValue"; 78 | 79 | case cudaErrorInvalidSymbol: 80 | return "cudaErrorInvalidSymbol"; 81 | 82 | case cudaErrorMapBufferObjectFailed: 83 | return "cudaErrorMapBufferObjectFailed"; 84 | 85 | case cudaErrorUnmapBufferObjectFailed: 86 | return "cudaErrorUnmapBufferObjectFailed"; 87 | 88 | case cudaErrorInvalidHostPointer: 89 | return "cudaErrorInvalidHostPointer"; 90 | 91 | case cudaErrorInvalidDevicePointer: 92 | return "cudaErrorInvalidDevicePointer"; 93 | 94 | case cudaErrorInvalidTexture: 95 | return "cudaErrorInvalidTexture"; 96 | 97 | case cudaErrorInvalidTextureBinding: 98 | return "cudaErrorInvalidTextureBinding"; 99 | 100 | case cudaErrorInvalidChannelDescriptor: 101 | return "cudaErrorInvalidChannelDescriptor"; 102 | 103 | case cudaErrorInvalidMemcpyDirection: 104 | return "cudaErrorInvalidMemcpyDirection"; 105 | 106 | case cudaErrorAddressOfConstant: 107 | return "cudaErrorAddressOfConstant"; 108 | 109 | case cudaErrorTextureFetchFailed: 110 | return "cudaErrorTextureFetchFailed"; 111 | 112 | case cudaErrorTextureNotBound: 113 | return "cudaErrorTextureNotBound"; 114 | 115 | case cudaErrorSynchronizationError: 116 | return "cudaErrorSynchronizationError"; 117 | 118 | case cudaErrorInvalidFilterSetting: 119 | return "cudaErrorInvalidFilterSetting"; 120 | 121 | case cudaErrorInvalidNormSetting: 122 | return "cudaErrorInvalidNormSetting"; 123 | 124 | case cudaErrorMixedDeviceExecution: 125 | return "cudaErrorMixedDeviceExecution"; 126 | 127 | case cudaErrorCudartUnloading: 128 | return "cudaErrorCudartUnloading"; 129 | 130 | case cudaErrorUnknown: 131 | return "cudaErrorUnknown"; 132 | 133 | case cudaErrorNotYetImplemented: 134 | return "cudaErrorNotYetImplemented"; 135 | 136 | case cudaErrorMemoryValueTooLarge: 137 | return "cudaErrorMemoryValueTooLarge"; 138 | 139 | case cudaErrorInvalidResourceHandle: 140 | return "cudaErrorInvalidResourceHandle"; 141 | 142 | case cudaErrorNotReady: 143 | return "cudaErrorNotReady"; 144 | 145 | case cudaErrorInsufficientDriver: 146 | return "cudaErrorInsufficientDriver"; 147 | 148 | case cudaErrorSetOnActiveProcess: 149 | return "cudaErrorSetOnActiveProcess"; 150 | 151 | case cudaErrorInvalidSurface: 152 | return "cudaErrorInvalidSurface"; 153 | 154 | case cudaErrorNoDevice: 155 | return "cudaErrorNoDevice"; 156 | 157 | case cudaErrorECCUncorrectable: 158 | return "cudaErrorECCUncorrectable"; 159 | 160 | case cudaErrorSharedObjectSymbolNotFound: 161 | return "cudaErrorSharedObjectSymbolNotFound"; 162 | 163 | case cudaErrorSharedObjectInitFailed: 164 | return "cudaErrorSharedObjectInitFailed"; 165 | 166 | case cudaErrorUnsupportedLimit: 167 | return "cudaErrorUnsupportedLimit"; 168 | 169 | case cudaErrorDuplicateVariableName: 170 | return "cudaErrorDuplicateVariableName"; 171 | 172 | case cudaErrorDuplicateTextureName: 173 | return "cudaErrorDuplicateTextureName"; 174 | 175 | case cudaErrorDuplicateSurfaceName: 176 | return "cudaErrorDuplicateSurfaceName"; 177 | 178 | case cudaErrorDevicesUnavailable: 179 | return "cudaErrorDevicesUnavailable"; 180 | 181 | case cudaErrorInvalidKernelImage: 182 | return "cudaErrorInvalidKernelImage"; 183 | 184 | case cudaErrorNoKernelImageForDevice: 185 | return "cudaErrorNoKernelImageForDevice"; 186 | 187 | case cudaErrorIncompatibleDriverContext: 188 | return "cudaErrorIncompatibleDriverContext"; 189 | 190 | case cudaErrorPeerAccessAlreadyEnabled: 191 | return "cudaErrorPeerAccessAlreadyEnabled"; 192 | 193 | case cudaErrorPeerAccessNotEnabled: 194 | return "cudaErrorPeerAccessNotEnabled"; 195 | 196 | case cudaErrorDeviceAlreadyInUse: 197 | return "cudaErrorDeviceAlreadyInUse"; 198 | 199 | case cudaErrorProfilerDisabled: 200 | return "cudaErrorProfilerDisabled"; 201 | 202 | case cudaErrorProfilerNotInitialized: 203 | return "cudaErrorProfilerNotInitialized"; 204 | 205 | case cudaErrorProfilerAlreadyStarted: 206 | return "cudaErrorProfilerAlreadyStarted"; 207 | 208 | case cudaErrorProfilerAlreadyStopped: 209 | return "cudaErrorProfilerAlreadyStopped"; 210 | 211 | #if __CUDA_API_VERSION >= 0x4000 212 | 213 | case cudaErrorAssert: 214 | return "cudaErrorAssert"; 215 | 216 | case cudaErrorTooManyPeers: 217 | return "cudaErrorTooManyPeers"; 218 | 219 | case cudaErrorHostMemoryAlreadyRegistered: 220 | return "cudaErrorHostMemoryAlreadyRegistered"; 221 | 222 | case cudaErrorHostMemoryNotRegistered: 223 | return "cudaErrorHostMemoryNotRegistered"; 224 | #endif 225 | 226 | case cudaErrorStartupFailure: 227 | return "cudaErrorStartupFailure"; 228 | 229 | case cudaErrorApiFailureBase: 230 | return "cudaErrorApiFailureBase"; 231 | } 232 | 233 | return ""; 234 | } 235 | #endif 236 | 237 | #ifdef __cuda_cuda_h__ 238 | // CUDA Driver API errors 239 | static const char *_cudaGetErrorEnum(CUresult error) 240 | { 241 | switch (error) 242 | { 243 | case CUDA_SUCCESS: 244 | return "CUDA_SUCCESS"; 245 | 246 | case CUDA_ERROR_INVALID_VALUE: 247 | return "CUDA_ERROR_INVALID_VALUE"; 248 | 249 | case CUDA_ERROR_OUT_OF_MEMORY: 250 | return "CUDA_ERROR_OUT_OF_MEMORY"; 251 | 252 | case CUDA_ERROR_NOT_INITIALIZED: 253 | return "CUDA_ERROR_NOT_INITIALIZED"; 254 | 255 | case CUDA_ERROR_DEINITIALIZED: 256 | return "CUDA_ERROR_DEINITIALIZED"; 257 | 258 | case CUDA_ERROR_PROFILER_DISABLED: 259 | return "CUDA_ERROR_PROFILER_DISABLED"; 260 | 261 | case CUDA_ERROR_PROFILER_NOT_INITIALIZED: 262 | return "CUDA_ERROR_PROFILER_NOT_INITIALIZED"; 263 | 264 | case CUDA_ERROR_PROFILER_ALREADY_STARTED: 265 | return "CUDA_ERROR_PROFILER_ALREADY_STARTED"; 266 | 267 | case CUDA_ERROR_PROFILER_ALREADY_STOPPED: 268 | return "CUDA_ERROR_PROFILER_ALREADY_STOPPED"; 269 | 270 | case CUDA_ERROR_NO_DEVICE: 271 | return "CUDA_ERROR_NO_DEVICE"; 272 | 273 | case CUDA_ERROR_INVALID_DEVICE: 274 | return "CUDA_ERROR_INVALID_DEVICE"; 275 | 276 | case CUDA_ERROR_INVALID_IMAGE: 277 | return "CUDA_ERROR_INVALID_IMAGE"; 278 | 279 | case CUDA_ERROR_INVALID_CONTEXT: 280 | return "CUDA_ERROR_INVALID_CONTEXT"; 281 | 282 | case CUDA_ERROR_CONTEXT_ALREADY_CURRENT: 283 | return "CUDA_ERROR_CONTEXT_ALREADY_CURRENT"; 284 | 285 | case CUDA_ERROR_MAP_FAILED: 286 | return "CUDA_ERROR_MAP_FAILED"; 287 | 288 | case CUDA_ERROR_UNMAP_FAILED: 289 | return "CUDA_ERROR_UNMAP_FAILED"; 290 | 291 | case CUDA_ERROR_ARRAY_IS_MAPPED: 292 | return "CUDA_ERROR_ARRAY_IS_MAPPED"; 293 | 294 | case CUDA_ERROR_ALREADY_MAPPED: 295 | return "CUDA_ERROR_ALREADY_MAPPED"; 296 | 297 | case CUDA_ERROR_NO_BINARY_FOR_GPU: 298 | return "CUDA_ERROR_NO_BINARY_FOR_GPU"; 299 | 300 | case CUDA_ERROR_ALREADY_ACQUIRED: 301 | return "CUDA_ERROR_ALREADY_ACQUIRED"; 302 | 303 | case CUDA_ERROR_NOT_MAPPED: 304 | return "CUDA_ERROR_NOT_MAPPED"; 305 | 306 | case CUDA_ERROR_NOT_MAPPED_AS_ARRAY: 307 | return "CUDA_ERROR_NOT_MAPPED_AS_ARRAY"; 308 | 309 | case CUDA_ERROR_NOT_MAPPED_AS_POINTER: 310 | return "CUDA_ERROR_NOT_MAPPED_AS_POINTER"; 311 | 312 | case CUDA_ERROR_ECC_UNCORRECTABLE: 313 | return "CUDA_ERROR_ECC_UNCORRECTABLE"; 314 | 315 | case CUDA_ERROR_UNSUPPORTED_LIMIT: 316 | return "CUDA_ERROR_UNSUPPORTED_LIMIT"; 317 | 318 | case CUDA_ERROR_CONTEXT_ALREADY_IN_USE: 319 | return "CUDA_ERROR_CONTEXT_ALREADY_IN_USE"; 320 | 321 | case CUDA_ERROR_INVALID_SOURCE: 322 | return "CUDA_ERROR_INVALID_SOURCE"; 323 | 324 | case CUDA_ERROR_FILE_NOT_FOUND: 325 | return "CUDA_ERROR_FILE_NOT_FOUND"; 326 | 327 | case CUDA_ERROR_SHARED_OBJECT_SYMBOL_NOT_FOUND: 328 | return "CUDA_ERROR_SHARED_OBJECT_SYMBOL_NOT_FOUND"; 329 | 330 | case CUDA_ERROR_SHARED_OBJECT_INIT_FAILED: 331 | return "CUDA_ERROR_SHARED_OBJECT_INIT_FAILED"; 332 | 333 | case CUDA_ERROR_OPERATING_SYSTEM: 334 | return "CUDA_ERROR_OPERATING_SYSTEM"; 335 | 336 | case CUDA_ERROR_INVALID_HANDLE: 337 | return "CUDA_ERROR_INVALID_HANDLE"; 338 | 339 | case CUDA_ERROR_NOT_FOUND: 340 | return "CUDA_ERROR_NOT_FOUND"; 341 | 342 | case CUDA_ERROR_NOT_READY: 343 | return "CUDA_ERROR_NOT_READY"; 344 | 345 | case CUDA_ERROR_LAUNCH_FAILED: 346 | return "CUDA_ERROR_LAUNCH_FAILED"; 347 | 348 | case CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES: 349 | return "CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES"; 350 | 351 | case CUDA_ERROR_LAUNCH_TIMEOUT: 352 | return "CUDA_ERROR_LAUNCH_TIMEOUT"; 353 | 354 | case CUDA_ERROR_LAUNCH_INCOMPATIBLE_TEXTURING: 355 | return "CUDA_ERROR_LAUNCH_INCOMPATIBLE_TEXTURING"; 356 | 357 | case CUDA_ERROR_PEER_ACCESS_ALREADY_ENABLED: 358 | return "CUDA_ERROR_PEER_ACCESS_ALREADY_ENABLED"; 359 | 360 | case CUDA_ERROR_PEER_ACCESS_NOT_ENABLED: 361 | return "CUDA_ERROR_PEER_ACCESS_NOT_ENABLED"; 362 | 363 | case CUDA_ERROR_PRIMARY_CONTEXT_ACTIVE: 364 | return "CUDA_ERROR_PRIMARY_CONTEXT_ACTIVE"; 365 | 366 | case CUDA_ERROR_CONTEXT_IS_DESTROYED: 367 | return "CUDA_ERROR_CONTEXT_IS_DESTROYED"; 368 | 369 | case CUDA_ERROR_ASSERT: 370 | return "CUDA_ERROR_ASSERT"; 371 | 372 | case CUDA_ERROR_TOO_MANY_PEERS: 373 | return "CUDA_ERROR_TOO_MANY_PEERS"; 374 | 375 | case CUDA_ERROR_HOST_MEMORY_ALREADY_REGISTERED: 376 | return "CUDA_ERROR_HOST_MEMORY_ALREADY_REGISTERED"; 377 | 378 | case CUDA_ERROR_HOST_MEMORY_NOT_REGISTERED: 379 | return "CUDA_ERROR_HOST_MEMORY_NOT_REGISTERED"; 380 | 381 | case CUDA_ERROR_UNKNOWN: 382 | return "CUDA_ERROR_UNKNOWN"; 383 | } 384 | 385 | return ""; 386 | } 387 | #endif 388 | 389 | #ifdef CUBLAS_API_H_ 390 | // cuBLAS API errors 391 | static const char *_cudaGetErrorEnum(cublasStatus_t error) 392 | { 393 | switch (error) 394 | { 395 | case CUBLAS_STATUS_SUCCESS: 396 | return "CUBLAS_STATUS_SUCCESS"; 397 | 398 | case CUBLAS_STATUS_NOT_INITIALIZED: 399 | return "CUBLAS_STATUS_NOT_INITIALIZED"; 400 | 401 | case CUBLAS_STATUS_ALLOC_FAILED: 402 | return "CUBLAS_STATUS_ALLOC_FAILED"; 403 | 404 | case CUBLAS_STATUS_INVALID_VALUE: 405 | return "CUBLAS_STATUS_INVALID_VALUE"; 406 | 407 | case CUBLAS_STATUS_ARCH_MISMATCH: 408 | return "CUBLAS_STATUS_ARCH_MISMATCH"; 409 | 410 | case CUBLAS_STATUS_MAPPING_ERROR: 411 | return "CUBLAS_STATUS_MAPPING_ERROR"; 412 | 413 | case CUBLAS_STATUS_EXECUTION_FAILED: 414 | return "CUBLAS_STATUS_EXECUTION_FAILED"; 415 | 416 | case CUBLAS_STATUS_INTERNAL_ERROR: 417 | return "CUBLAS_STATUS_INTERNAL_ERROR"; 418 | } 419 | 420 | return ""; 421 | } 422 | #endif 423 | 424 | #ifdef _CUFFT_H_ 425 | // cuFFT API errors 426 | static const char *_cudaGetErrorEnum(cufftResult error) 427 | { 428 | switch (error) 429 | { 430 | case CUFFT_SUCCESS: 431 | return "CUFFT_SUCCESS"; 432 | 433 | case CUFFT_INVALID_PLAN: 434 | return "CUFFT_INVALID_PLAN"; 435 | 436 | case CUFFT_ALLOC_FAILED: 437 | return "CUFFT_ALLOC_FAILED"; 438 | 439 | case CUFFT_INVALID_TYPE: 440 | return "CUFFT_INVALID_TYPE"; 441 | 442 | case CUFFT_INVALID_VALUE: 443 | return "CUFFT_INVALID_VALUE"; 444 | 445 | case CUFFT_INTERNAL_ERROR: 446 | return "CUFFT_INTERNAL_ERROR"; 447 | 448 | case CUFFT_EXEC_FAILED: 449 | return "CUFFT_EXEC_FAILED"; 450 | 451 | case CUFFT_SETUP_FAILED: 452 | return "CUFFT_SETUP_FAILED"; 453 | 454 | case CUFFT_INVALID_SIZE: 455 | return "CUFFT_INVALID_SIZE"; 456 | 457 | case CUFFT_UNALIGNED_DATA: 458 | return "CUFFT_UNALIGNED_DATA"; 459 | } 460 | 461 | return ""; 462 | } 463 | #endif 464 | 465 | 466 | #ifdef CUSPARSEAPI 467 | // cuSPARSE API errors 468 | static const char *_cudaGetErrorEnum(cusparseStatus_t error) 469 | { 470 | switch (error) 471 | { 472 | case CUSPARSE_STATUS_SUCCESS: 473 | return "CUSPARSE_STATUS_SUCCESS"; 474 | 475 | case CUSPARSE_STATUS_NOT_INITIALIZED: 476 | return "CUSPARSE_STATUS_NOT_INITIALIZED"; 477 | 478 | case CUSPARSE_STATUS_ALLOC_FAILED: 479 | return "CUSPARSE_STATUS_ALLOC_FAILED"; 480 | 481 | case CUSPARSE_STATUS_INVALID_VALUE: 482 | return "CUSPARSE_STATUS_INVALID_VALUE"; 483 | 484 | case CUSPARSE_STATUS_ARCH_MISMATCH: 485 | return "CUSPARSE_STATUS_ARCH_MISMATCH"; 486 | 487 | case CUSPARSE_STATUS_MAPPING_ERROR: 488 | return "CUSPARSE_STATUS_MAPPING_ERROR"; 489 | 490 | case CUSPARSE_STATUS_EXECUTION_FAILED: 491 | return "CUSPARSE_STATUS_EXECUTION_FAILED"; 492 | 493 | case CUSPARSE_STATUS_INTERNAL_ERROR: 494 | return "CUSPARSE_STATUS_INTERNAL_ERROR"; 495 | 496 | case CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED: 497 | return "CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED"; 498 | } 499 | 500 | return ""; 501 | } 502 | #endif 503 | 504 | #ifdef CURAND_H_ 505 | // cuRAND API errors 506 | static const char *_cudaGetErrorEnum(curandStatus_t error) 507 | { 508 | switch (error) 509 | { 510 | case CURAND_STATUS_SUCCESS: 511 | return "CURAND_STATUS_SUCCESS"; 512 | 513 | case CURAND_STATUS_VERSION_MISMATCH: 514 | return "CURAND_STATUS_VERSION_MISMATCH"; 515 | 516 | case CURAND_STATUS_NOT_INITIALIZED: 517 | return "CURAND_STATUS_NOT_INITIALIZED"; 518 | 519 | case CURAND_STATUS_ALLOCATION_FAILED: 520 | return "CURAND_STATUS_ALLOCATION_FAILED"; 521 | 522 | case CURAND_STATUS_TYPE_ERROR: 523 | return "CURAND_STATUS_TYPE_ERROR"; 524 | 525 | case CURAND_STATUS_OUT_OF_RANGE: 526 | return "CURAND_STATUS_OUT_OF_RANGE"; 527 | 528 | case CURAND_STATUS_LENGTH_NOT_MULTIPLE: 529 | return "CURAND_STATUS_LENGTH_NOT_MULTIPLE"; 530 | 531 | case CURAND_STATUS_DOUBLE_PRECISION_REQUIRED: 532 | return "CURAND_STATUS_DOUBLE_PRECISION_REQUIRED"; 533 | 534 | case CURAND_STATUS_LAUNCH_FAILURE: 535 | return "CURAND_STATUS_LAUNCH_FAILURE"; 536 | 537 | case CURAND_STATUS_PREEXISTING_FAILURE: 538 | return "CURAND_STATUS_PREEXISTING_FAILURE"; 539 | 540 | case CURAND_STATUS_INITIALIZATION_FAILED: 541 | return "CURAND_STATUS_INITIALIZATION_FAILED"; 542 | 543 | case CURAND_STATUS_ARCH_MISMATCH: 544 | return "CURAND_STATUS_ARCH_MISMATCH"; 545 | 546 | case CURAND_STATUS_INTERNAL_ERROR: 547 | return "CURAND_STATUS_INTERNAL_ERROR"; 548 | } 549 | 550 | return ""; 551 | } 552 | #endif 553 | 554 | #ifdef NV_NPPIDEFS_H 555 | // NPP API errors 556 | static const char *_cudaGetErrorEnum(NppStatus error) 557 | { 558 | switch (error) 559 | { 560 | case NPP_NOT_SUPPORTED_MODE_ERROR: 561 | return "NPP_NOT_SUPPORTED_MODE_ERROR"; 562 | 563 | case NPP_ROUND_MODE_NOT_SUPPORTED_ERROR: 564 | return "NPP_ROUND_MODE_NOT_SUPPORTED_ERROR"; 565 | 566 | case NPP_RESIZE_NO_OPERATION_ERROR: 567 | return "NPP_RESIZE_NO_OPERATION_ERROR"; 568 | 569 | case NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY: 570 | return "NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY"; 571 | 572 | case NPP_BAD_ARG_ERROR: 573 | return "NPP_BAD_ARG_ERROR"; 574 | 575 | case NPP_LUT_NUMBER_OF_LEVELS_ERROR: 576 | return "NPP_LUT_NUMBER_OF_LEVELS_ERROR"; 577 | 578 | case NPP_TEXTURE_BIND_ERROR: 579 | return "NPP_TEXTURE_BIND_ERROR"; 580 | 581 | case NPP_COEFF_ERROR: 582 | return "NPP_COEFF_ERROR"; 583 | 584 | case NPP_RECT_ERROR: 585 | return "NPP_RECT_ERROR"; 586 | 587 | case NPP_QUAD_ERROR: 588 | return "NPP_QUAD_ERROR"; 589 | 590 | case NPP_WRONG_INTERSECTION_ROI_ERROR: 591 | return "NPP_WRONG_INTERSECTION_ROI_ERROR"; 592 | 593 | case NPP_NOT_EVEN_STEP_ERROR: 594 | return "NPP_NOT_EVEN_STEP_ERROR"; 595 | 596 | case NPP_INTERPOLATION_ERROR: 597 | return "NPP_INTERPOLATION_ERROR"; 598 | 599 | case NPP_RESIZE_FACTOR_ERROR: 600 | return "NPP_RESIZE_FACTOR_ERROR"; 601 | 602 | case NPP_HAAR_CLASSIFIER_PIXEL_MATCH_ERROR: 603 | return "NPP_HAAR_CLASSIFIER_PIXEL_MATCH_ERROR"; 604 | 605 | case NPP_MEMFREE_ERR: 606 | return "NPP_MEMFREE_ERR"; 607 | 608 | case NPP_MEMSET_ERR: 609 | return "NPP_MEMSET_ERR"; 610 | 611 | case NPP_MEMCPY_ERROR: 612 | return "NPP_MEMCPY_ERROR"; 613 | 614 | case NPP_MEM_ALLOC_ERR: 615 | return "NPP_MEM_ALLOC_ERR"; 616 | 617 | case NPP_HISTO_NUMBER_OF_LEVELS_ERROR: 618 | return "NPP_HISTO_NUMBER_OF_LEVELS_ERROR"; 619 | 620 | case NPP_MIRROR_FLIP_ERR: 621 | return "NPP_MIRROR_FLIP_ERR"; 622 | 623 | case NPP_INVALID_INPUT: 624 | return "NPP_INVALID_INPUT"; 625 | 626 | case NPP_ALIGNMENT_ERROR: 627 | return "NPP_ALIGNMENT_ERROR"; 628 | 629 | case NPP_STEP_ERROR: 630 | return "NPP_STEP_ERROR"; 631 | 632 | case NPP_SIZE_ERROR: 633 | return "NPP_SIZE_ERROR"; 634 | 635 | case NPP_POINTER_ERROR: 636 | return "NPP_POINTER_ERROR"; 637 | 638 | case NPP_NULL_POINTER_ERROR: 639 | return "NPP_NULL_POINTER_ERROR"; 640 | 641 | case NPP_CUDA_KERNEL_EXECUTION_ERROR: 642 | return "NPP_CUDA_KERNEL_EXECUTION_ERROR"; 643 | 644 | case NPP_NOT_IMPLEMENTED_ERROR: 645 | return "NPP_NOT_IMPLEMENTED_ERROR"; 646 | 647 | case NPP_ERROR: 648 | return "NPP_ERROR"; 649 | 650 | case NPP_SUCCESS: 651 | return "NPP_SUCCESS"; 652 | 653 | case NPP_WARNING: 654 | return "NPP_WARNING"; 655 | 656 | case NPP_WRONG_INTERSECTION_QUAD_WARNING: 657 | return "NPP_WRONG_INTERSECTION_QUAD_WARNING"; 658 | 659 | case NPP_MISALIGNED_DST_ROI_WARNING: 660 | return "NPP_MISALIGNED_DST_ROI_WARNING"; 661 | 662 | case NPP_AFFINE_QUAD_INCORRECT_WARNING: 663 | return "NPP_AFFINE_QUAD_INCORRECT_WARNING"; 664 | 665 | case NPP_DOUBLE_SIZE_WARNING: 666 | return "NPP_DOUBLE_SIZE_WARNING"; 667 | 668 | case NPP_ODD_ROI_WARNING: 669 | return "NPP_ODD_ROI_WARNING"; 670 | 671 | case NPP_WRONG_INTERSECTION_ROI_WARNING: 672 | return "NPP_WRONG_INTERSECTION_ROI_WARNING"; 673 | } 674 | 675 | return ""; 676 | } 677 | #endif 678 | 679 | template< typename T > 680 | bool check(T result, char const *const func, const char *const file, int const line) 681 | { 682 | if (result) 683 | { 684 | fprintf(stderr, "CUDA error at %s:%d code=%d(%s) \"%s\" \n", 685 | file, line, static_cast(result), _cudaGetErrorEnum(result), func); 686 | /* 687 | std::stringstream ss; 688 | std::string msg("CUDA error at "); 689 | msg += file; 690 | msg += ":"; 691 | ss << line; 692 | msg += ss.str(); 693 | msg += " code="; 694 | ss << static_cast(result); 695 | msg += ss.str(); 696 | msg += " ("; 697 | msg += _cudaGetErrorEnum(result); 698 | msg += ") \""; 699 | msg += func; 700 | msg += "\""; 701 | //throw msg; 702 | std::cerr << msg <<"\n"; 703 | */ 704 | return true; 705 | } 706 | else 707 | { 708 | return false; 709 | } 710 | } 711 | 712 | #ifdef __DRIVER_TYPES_H__ 713 | // This will output the proper CUDA error strings in the event that a CUDA host call returns an error 714 | #define checkCudaErrors(val) check ( (val), #val, __FILE__, __LINE__ ) 715 | 716 | // This will output the proper error string when calling cudaGetLastError 717 | #define getLastCudaError(msg) __getLastCudaError (msg, __FILE__, __LINE__) 718 | 719 | inline void __getLastCudaError(const char *errorMessage, const char *file, const int line) 720 | { 721 | cudaError_t err = cudaGetLastError(); 722 | 723 | if (cudaSuccess != err) 724 | { 725 | fprintf(stderr, "%s(%i) : getLastCudaError() CUDA error : %s : (%d) %s.\n", 726 | file, line, errorMessage, (int)err, cudaGetErrorString(err)); 727 | exit(EXIT_FAILURE); 728 | } 729 | } 730 | #endif 731 | 732 | #ifndef MAX 733 | #define MAX(a,b) (a > b ? a : b) 734 | #endif 735 | 736 | // Beginning of GPU Architecture definitions 737 | inline int _ConvertSMVer2Cores(int major, int minor) 738 | { 739 | // Defines for GPU Architecture types (using the SM version to determine the # of cores per SM 740 | typedef struct 741 | { 742 | int SM; // 0xMm (hexidecimal notation), M = SM Major version, and m = SM minor version 743 | int Cores; 744 | } sSMtoCores; 745 | 746 | sSMtoCores nGpuArchCoresPerSM[] = 747 | { 748 | { 0x10, 8 }, // Tesla Generation (SM 1.0) G80 class 749 | { 0x11, 8 }, // Tesla Generation (SM 1.1) G8x class 750 | { 0x12, 8 }, // Tesla Generation (SM 1.2) G9x class 751 | { 0x13, 8 }, // Tesla Generation (SM 1.3) GT200 class 752 | { 0x20, 32 }, // Fermi Generation (SM 2.0) GF100 class 753 | { 0x21, 48 }, // Fermi Generation (SM 2.1) GF10x class 754 | { 0x30, 192}, // Kepler Generation (SM 3.0) GK10x class 755 | { 0x35, 192}, // Kepler Generation (SM 3.5) GK11x class 756 | { -1, -1 } 757 | }; 758 | 759 | int index = 0; 760 | 761 | while (nGpuArchCoresPerSM[index].SM != -1) 762 | { 763 | if (nGpuArchCoresPerSM[index].SM == ((major << 4) + minor)) 764 | { 765 | return nGpuArchCoresPerSM[index].Cores; 766 | } 767 | 768 | index++; 769 | } 770 | 771 | // If we don't find the values, we default use the previous one to run properly 772 | printf("MapSMtoCores for SM %d.%d is undefined. Default to use %d Cores/SM\n", major, minor, nGpuArchCoresPerSM[7].Cores); 773 | return nGpuArchCoresPerSM[7].Cores; 774 | } 775 | // end of GPU Architecture definitions 776 | 777 | #ifdef __CUDA_RUNTIME_H__ 778 | // General GPU Device CUDA Initialization 779 | inline int gpuDeviceInit(int devID) 780 | { 781 | int deviceCount; 782 | checkCudaErrors(cudaGetDeviceCount(&deviceCount)); 783 | 784 | if (deviceCount == 0) 785 | { 786 | fprintf(stderr, "gpuDeviceInit() CUDA error: no devices supporting CUDA.\n"); 787 | exit(EXIT_FAILURE); 788 | } 789 | 790 | if (devID < 0) 791 | { 792 | devID = 0; 793 | } 794 | 795 | if (devID > deviceCount-1) 796 | { 797 | fprintf(stderr, "\n"); 798 | fprintf(stderr, ">> %d CUDA capable GPU device(s) detected. <<\n", deviceCount); 799 | fprintf(stderr, ">> gpuDeviceInit (-device=%d) is not a valid GPU device. <<\n", devID); 800 | fprintf(stderr, "\n"); 801 | return -devID; 802 | } 803 | 804 | cudaDeviceProp deviceProp; 805 | checkCudaErrors(cudaGetDeviceProperties(&deviceProp, devID)); 806 | 807 | if (deviceProp.computeMode == cudaComputeModeProhibited) 808 | { 809 | fprintf(stderr, "Error: device is running in , no threads can use ::cudaSetDevice().\n"); 810 | return -1; 811 | } 812 | 813 | if (deviceProp.major < 1) 814 | { 815 | fprintf(stderr, "gpuDeviceInit(): GPU device does not support CUDA.\n"); 816 | exit(EXIT_FAILURE); 817 | } 818 | 819 | checkCudaErrors(cudaSetDevice(devID)); 820 | printf("gpuDeviceInit() CUDA Device [%d]: \"%s\n", devID, deviceProp.name); 821 | 822 | return devID; 823 | } 824 | 825 | // This function returns the best GPU (with maximum GFLOPS) 826 | inline int gpuGetMaxGflopsDeviceId() 827 | { 828 | int current_device = 0, sm_per_multiproc = 0; 829 | int max_compute_perf = 0, max_perf_device = 0; 830 | int device_count = 0, best_SM_arch = 0; 831 | cudaDeviceProp deviceProp; 832 | cudaGetDeviceCount(&device_count); 833 | 834 | // Find the best major SM Architecture GPU device 835 | while (current_device < device_count) 836 | { 837 | cudaGetDeviceProperties(&deviceProp, current_device); 838 | 839 | // If this GPU is not running on Compute Mode prohibited, then we can add it to the list 840 | if (deviceProp.computeMode != cudaComputeModeProhibited) 841 | { 842 | if (deviceProp.major > 0 && deviceProp.major < 9999) 843 | { 844 | best_SM_arch = MAX(best_SM_arch, deviceProp.major); 845 | } 846 | } 847 | 848 | current_device++; 849 | } 850 | 851 | // Find the best CUDA capable GPU device 852 | current_device = 0; 853 | 854 | while (current_device < device_count) 855 | { 856 | cudaGetDeviceProperties(&deviceProp, current_device); 857 | 858 | // If this GPU is not running on Compute Mode prohibited, then we can add it to the list 859 | if (deviceProp.computeMode != cudaComputeModeProhibited) 860 | { 861 | if (deviceProp.major == 9999 && deviceProp.minor == 9999) 862 | { 863 | sm_per_multiproc = 1; 864 | } 865 | else 866 | { 867 | sm_per_multiproc = _ConvertSMVer2Cores(deviceProp.major, deviceProp.minor); 868 | } 869 | 870 | int compute_perf = deviceProp.multiProcessorCount * sm_per_multiproc * deviceProp.clockRate; 871 | 872 | if (compute_perf > max_compute_perf) 873 | { 874 | // If we find GPU with SM major > 2, search only these 875 | if (best_SM_arch > 2) 876 | { 877 | // If our device==dest_SM_arch, choose this, or else pass 878 | if (deviceProp.major == best_SM_arch) 879 | { 880 | max_compute_perf = compute_perf; 881 | max_perf_device = current_device; 882 | } 883 | } 884 | else 885 | { 886 | max_compute_perf = compute_perf; 887 | max_perf_device = current_device; 888 | } 889 | } 890 | } 891 | 892 | ++current_device; 893 | } 894 | 895 | return max_perf_device; 896 | } 897 | 898 | 899 | // Initialization code to find the best CUDA Device 900 | inline int findCudaDevice(int argc, const char **argv) 901 | { 902 | cudaDeviceProp deviceProp; 903 | int devID = 0; 904 | 905 | // If the command-line has a device number specified, use it 906 | if (checkCmdLineFlag(argc, argv, "device")) 907 | { 908 | devID = getCmdLineArgumentInt(argc, argv, "device="); 909 | 910 | if (devID < 0) 911 | { 912 | printf("Invalid command line parameter\n "); 913 | exit(EXIT_FAILURE); 914 | } 915 | else 916 | { 917 | devID = gpuDeviceInit(devID); 918 | 919 | if (devID < 0) 920 | { 921 | printf("exiting...\n"); 922 | exit(EXIT_FAILURE); 923 | } 924 | } 925 | } 926 | else 927 | { 928 | // Otherwise pick the device with highest Gflops/s 929 | devID = gpuGetMaxGflopsDeviceId(); 930 | checkCudaErrors(cudaSetDevice(devID)); 931 | checkCudaErrors(cudaGetDeviceProperties(&deviceProp, devID)); 932 | printf("GPU Device %d: \"%s\" with compute capability %d.%d\n\n", devID, deviceProp.name, deviceProp.major, deviceProp.minor); 933 | } 934 | 935 | return devID; 936 | } 937 | 938 | // General check for CUDA GPU SM Capabilities 939 | inline bool checkCudaCapabilities(int major_version, int minor_version) 940 | { 941 | cudaDeviceProp deviceProp; 942 | deviceProp.major = 0; 943 | deviceProp.minor = 0; 944 | int dev; 945 | 946 | checkCudaErrors(cudaGetDevice(&dev)); 947 | checkCudaErrors(cudaGetDeviceProperties(&deviceProp, dev)); 948 | 949 | if ((deviceProp.major > major_version) || 950 | (deviceProp.major == major_version && deviceProp.minor >= minor_version)) 951 | { 952 | printf("> Device %d: <%16s >, Compute SM %d.%d detected\n", dev, deviceProp.name, deviceProp.major, deviceProp.minor); 953 | return true; 954 | } 955 | else 956 | { 957 | printf("No GPU device was found that can support CUDA compute capability %d.%d.\n", major_version, minor_version); 958 | return false; 959 | } 960 | } 961 | #endif 962 | 963 | // end of CUDA Helper Functions 964 | 965 | 966 | #endif -------------------------------------------------------------------------------- /adaptive_gridsampler/helper_string.h: -------------------------------------------------------------------------------- 1 | /** 2 | * Copyright 1993-2012 NVIDIA Corporation. All rights reserved. 3 | * 4 | * Please refer to the NVIDIA end user license agreement (EULA) associated 5 | * with this source code for terms and conditions that govern your use of 6 | * this software. Any use, reproduction, disclosure, or distribution of 7 | * this software and related documentation outside the terms of the EULA 8 | * is strictly prohibited. 9 | * 10 | */ 11 | 12 | // These are helper functions for the SDK samples (string parsing, timers, etc) 13 | #ifndef STRING_HELPER_H 14 | #define STRING_HELPER_H 15 | 16 | #include 17 | #include 18 | #include 19 | #include 20 | 21 | #ifdef _WIN32 22 | #ifndef STRCASECMP 23 | #define STRCASECMP _stricmp 24 | #endif 25 | #ifndef STRNCASECMP 26 | #define STRNCASECMP _strnicmp 27 | #endif 28 | #ifndef STRCPY 29 | #define STRCPY(sFilePath, nLength, sPath) strcpy_s(sFilePath, nLength, sPath) 30 | #endif 31 | 32 | #ifndef FOPEN 33 | #define FOPEN(fHandle,filename,mode) fopen_s(&fHandle, filename, mode) 34 | #endif 35 | #ifndef FOPEN_FAIL 36 | #define FOPEN_FAIL(result) (result != 0) 37 | #endif 38 | #ifndef SSCANF 39 | #define SSCANF sscanf_s 40 | #endif 41 | 42 | #else 43 | #include 44 | #include 45 | 46 | #ifndef STRCASECMP 47 | #define STRCASECMP strcasecmp 48 | #endif 49 | #ifndef STRNCASECMP 50 | #define STRNCASECMP strncasecmp 51 | #endif 52 | #ifndef STRCPY 53 | #define STRCPY(sFilePath, nLength, sPath) strcpy(sFilePath, sPath) 54 | #endif 55 | 56 | #ifndef FOPEN 57 | #define FOPEN(fHandle,filename,mode) (fHandle = fopen(filename, mode)) 58 | #endif 59 | #ifndef FOPEN_FAIL 60 | #define FOPEN_FAIL(result) (result == NULL) 61 | #endif 62 | #ifndef SSCANF 63 | #define SSCANF sscanf 64 | #endif 65 | #endif 66 | 67 | // CUDA Utility Helper Functions 68 | inline int stringRemoveDelimiter(char delimiter, const char *string) 69 | { 70 | int string_start = 0; 71 | 72 | while (string[string_start] == delimiter) 73 | { 74 | string_start++; 75 | } 76 | 77 | if (string_start >= (int)strlen(string)-1) 78 | { 79 | return 0; 80 | } 81 | 82 | return string_start; 83 | } 84 | 85 | inline int getFileExtension(char *filename, char **extension) 86 | { 87 | int string_length = (int)strlen(filename); 88 | 89 | while (filename[string_length--] != '.') { 90 | if (string_length == 0) 91 | break; 92 | } 93 | if (string_length > 0) string_length += 2; 94 | 95 | if (string_length == 0) 96 | *extension = NULL; 97 | else 98 | *extension = &filename[string_length]; 99 | 100 | return string_length; 101 | } 102 | 103 | 104 | inline int checkCmdLineFlag(const int argc, const char **argv, const char *string_ref) 105 | { 106 | bool bFound = false; 107 | 108 | if (argc >= 1) 109 | { 110 | for (int i=1; i < argc; i++) 111 | { 112 | int string_start = stringRemoveDelimiter('-', argv[i]); 113 | const char *string_argv = &argv[i][string_start]; 114 | 115 | const char *equal_pos = strchr(string_argv, '='); 116 | int argv_length = (int)(equal_pos == 0 ? strlen(string_argv) : equal_pos - string_argv); 117 | 118 | int length = (int)strlen(string_ref); 119 | 120 | if (length == argv_length && !STRNCASECMP(string_argv, string_ref, length)) 121 | { 122 | 123 | bFound = true; 124 | continue; 125 | } 126 | } 127 | } 128 | 129 | return (int)bFound; 130 | } 131 | 132 | inline int getCmdLineArgumentInt(const int argc, const char **argv, const char *string_ref) 133 | { 134 | bool bFound = false; 135 | int value = -1; 136 | 137 | if (argc >= 1) 138 | { 139 | for (int i=1; i < argc; i++) 140 | { 141 | int string_start = stringRemoveDelimiter('-', argv[i]); 142 | const char *string_argv = &argv[i][string_start]; 143 | int length = (int)strlen(string_ref); 144 | 145 | if (!STRNCASECMP(string_argv, string_ref, length)) 146 | { 147 | if (length+1 <= (int)strlen(string_argv)) 148 | { 149 | int auto_inc = (string_argv[length] == '=') ? 1 : 0; 150 | value = atoi(&string_argv[length + auto_inc]); 151 | } 152 | else 153 | { 154 | value = 0; 155 | } 156 | 157 | bFound = true; 158 | continue; 159 | } 160 | } 161 | } 162 | 163 | if (bFound) 164 | { 165 | return value; 166 | } 167 | else 168 | { 169 | return 0; 170 | } 171 | } 172 | 173 | inline float getCmdLineArgumentFloat(const int argc, const char **argv, const char *string_ref) 174 | { 175 | bool bFound = false; 176 | float value = -1; 177 | 178 | if (argc >= 1) 179 | { 180 | for (int i=1; i < argc; i++) 181 | { 182 | int string_start = stringRemoveDelimiter('-', argv[i]); 183 | const char *string_argv = &argv[i][string_start]; 184 | int length = (int)strlen(string_ref); 185 | 186 | if (!STRNCASECMP(string_argv, string_ref, length)) 187 | { 188 | if (length+1 <= (int)strlen(string_argv)) 189 | { 190 | int auto_inc = (string_argv[length] == '=') ? 1 : 0; 191 | value = (float)atof(&string_argv[length + auto_inc]); 192 | } 193 | else 194 | { 195 | value = 0.f; 196 | } 197 | 198 | bFound = true; 199 | continue; 200 | } 201 | } 202 | } 203 | 204 | if (bFound) 205 | { 206 | return value; 207 | } 208 | else 209 | { 210 | return 0; 211 | } 212 | } 213 | 214 | inline bool getCmdLineArgumentString(const int argc, const char **argv, 215 | const char *string_ref, char **string_retval) 216 | { 217 | bool bFound = false; 218 | 219 | if (argc >= 1) 220 | { 221 | for (int i=1; i < argc; i++) 222 | { 223 | int string_start = stringRemoveDelimiter('-', argv[i]); 224 | char *string_argv = (char *)&argv[i][string_start]; 225 | int length = (int)strlen(string_ref); 226 | 227 | if (!STRNCASECMP(string_argv, string_ref, length)) 228 | { 229 | *string_retval = &string_argv[length+1]; 230 | bFound = true; 231 | continue; 232 | } 233 | } 234 | } 235 | 236 | if (!bFound) 237 | { 238 | *string_retval = NULL; 239 | } 240 | 241 | return bFound; 242 | } 243 | 244 | ////////////////////////////////////////////////////////////////////////////// 245 | //! Find the path for a file assuming that 246 | //! files are found in the searchPath. 247 | //! 248 | //! @return the path if succeeded, otherwise 0 249 | //! @param filename name of the file 250 | //! @param executable_path optional absolute path of the executable 251 | ////////////////////////////////////////////////////////////////////////////// 252 | inline char *sdkFindFilePath(const char *filename, const char *executable_path) 253 | { 254 | // defines a variable that is replaced with the name of the executable 255 | 256 | // Typical relative search paths to locate needed companion files (e.g. sample input data, or JIT source files) 257 | // The origin for the relative search may be the .exe file, a .bat file launching an .exe, a browser .exe launching the .exe or .bat, etc 258 | const char *searchPath[] = 259 | { 260 | "./", // same dir 261 | "./common/", // "/common/" subdir 262 | "./common/data/", // "/common/data/" subdir 263 | "./data/", // "/data/" subdir 264 | "./src/", // "/src/" subdir 265 | "./src//data/", // "/src//data/" subdir 266 | "./inc/", // "/inc/" subdir 267 | "./0_Simple/", // "/0_Simple/" subdir 268 | "./1_Utilities/", // "/1_Utilities/" subdir 269 | "./2_Graphics/", // "/2_Graphics/" subdir 270 | "./3_Imaging/", // "/3_Imaging/" subdir 271 | "./4_Financial/", // "/4_Financial/" subdir 272 | "./5_Simulations/", // "/5_Simulations/" subdir 273 | "./6_Advanced/", // "/6_Advanced/" subdir 274 | "./7_CUDALibraries/", // "/7_CUDALibraries/" subdir 275 | 276 | "../", // up 1 in tree 277 | "../common/", // up 1 in tree, "/common/" subdir 278 | "../common/data/", // up 1 in tree, "/common/data/" subdir 279 | "../data/", // up 1 in tree, "/data/" subdir 280 | "../src/", // up 1 in tree, "/src/" subdir 281 | "../inc/", // up 1 in tree, "/inc/" subdir 282 | "../C/src//", // up 1 in tree, "/C/src//" subdir 283 | "../C/src//data/", // up 1 in tree, "/C/src//data/" subdir 284 | "../C/src//src/", // up 1 in tree, "/C/src//src/" subdir 285 | "../C/src//inc/", // up 1 in tree, "/C/src//inc/" subdir 286 | "../C/", // up 1 in tree 287 | "../C/common/", // up 1 in tree, "/common/" subdir 288 | "../C/common/data/", // up 1 in tree, "/common/data/" subdir 289 | "../C/data/", // up 1 in tree, "/data/" subdir 290 | "../C/src/", // up 1 in tree, "/src/" subdir 291 | "../C/inc/", // up 1 in tree, "/inc/" subdir 292 | "../C/0_Simple//data/", // up 1 in tree, "/0_Simple//" subdir 293 | "../C/1_Utilities//data/", // up 1 in tree, "/1_Utilities//" subdir 294 | "../C/2_Graphics//data/", // up 1 in tree, "/2_Graphics//" subdir 295 | "../C/3_Imaging//data/", // up 1 in tree, "/3_Imaging//" subdir 296 | "../C/4_Financial//data/", // up 1 in tree, "/4_Financial//" subdir 297 | "../C/5_Simulations//data/", // up 1 in tree, "/5_Simulations//" subdir 298 | "../C/6_Advanced//data/", // up 1 in tree, "/6_Advanced//" subdir 299 | "../C/7_CUDALibraries//data/", // up 1 in tree, "/7_CUDALibraries//" subdir 300 | 301 | "../0_Simple//data/", // up 1 in tree, "/0_Simple//" subdir 302 | "../1_Utilities//data/", // up 1 in tree, "/1_Utilities//" subdir 303 | "../2_Graphics//data/", // up 1 in tree, "/2_Graphics//" subdir 304 | "../3_Imaging//data/", // up 1 in tree, "/3_Imaging//" subdir 305 | "../4_Financial//data/", // up 1 in tree, "/4_Financial//" subdir 306 | "../5_Simulations//data/", // up 1 in tree, "/5_Simulations//" subdir 307 | "../6_Advanced//data/", // up 1 in tree, "/6_Advanced//" subdir 308 | "../7_CUDALibraries//data/", // up 1 in tree, "/7_CUDALibraries//" subdir 309 | "../../", // up 2 in tree 310 | "../../common/", // up 2 in tree, "/common/" subdir 311 | "../../common/data/", // up 2 in tree, "/common/data/" subdir 312 | "../../data/", // up 2 in tree, "/data/" subdir 313 | "../../src/", // up 2 in tree, "/src/" subdir 314 | "../../inc/", // up 2 in tree, "/inc/" subdir 315 | "../../sandbox//data/", // up 2 in tree, "/sandbox//" subdir 316 | "../../0_Simple//data/", // up 2 in tree, "/0_Simple//" subdir 317 | "../../1_Utilities//data/", // up 2 in tree, "/1_Utilities//" subdir 318 | "../../2_Graphics//data/", // up 2 in tree, "/2_Graphics//" subdir 319 | "../../3_Imaging//data/", // up 2 in tree, "/3_Imaging//" subdir 320 | "../../4_Financial//data/", // up 2 in tree, "/4_Financial//" subdir 321 | "../../5_Simulations//data/", // up 2 in tree, "/5_Simulations//" subdir 322 | "../../6_Advanced//data/", // up 2 in tree, "/6_Advanced//" subdir 323 | "../../7_CUDALibraries//data/", // up 2 in tree, "/7_CUDALibraries//" subdir 324 | "../../../", // up 3 in tree 325 | "../../../src//", // up 3 in tree, "/src//" subdir 326 | "../../../src//data/", // up 3 in tree, "/src//data/" subdir 327 | "../../../src//src/", // up 3 in tree, "/src//src/" subdir 328 | "../../../src//inc/", // up 3 in tree, "/src//inc/" subdir 329 | "../../../sandbox//", // up 3 in tree, "/sandbox//" subdir 330 | "../../../sandbox//data/", // up 3 in tree, "/sandbox//data/" subdir 331 | "../../../sandbox//src/", // up 3 in tree, "/sandbox//src/" subdir 332 | "../../../sandbox//inc/", // up 3 in tree, "/sandbox//inc/" subdir 333 | "../../../0_Simple//data/", // up 3 in tree, "/0_Simple//" subdir 334 | "../../../1_Utilities//data/", // up 3 in tree, "/1_Utilities//" subdir 335 | "../../../2_Graphics//data/", // up 3 in tree, "/2_Graphics//" subdir 336 | "../../../3_Imaging//data/", // up 3 in tree, "/3_Imaging//" subdir 337 | "../../../4_Financial//data/", // up 3 in tree, "/4_Financial//" subdir 338 | "../../../5_Simulations//data/",// up 3 in tree, "/5_Simulations//" subdir 339 | "../../../6_Advanced//data/", // up 3 in tree, "/6_Advanced//" subdir 340 | "../../../7_CUDALibraries//data/", // up 3 in tree, "/7_CUDALibraries//" subdir 341 | "../../../common/", // up 3 in tree, "../../../common/" subdir 342 | "../../../common/data/", // up 3 in tree, "../../../common/data/" subdir 343 | "../../../data/", // up 3 in tree, "../../../data/" subdir 344 | }; 345 | 346 | // Extract the executable name 347 | std::string executable_name; 348 | 349 | if (executable_path != 0) 350 | { 351 | executable_name = std::string(executable_path); 352 | 353 | #ifdef _WIN32 354 | // Windows path delimiter 355 | size_t delimiter_pos = executable_name.find_last_of('\\'); 356 | executable_name.erase(0, delimiter_pos + 1); 357 | 358 | if (executable_name.rfind(".exe") != std::string::npos) 359 | { 360 | // we strip .exe, only if the .exe is found 361 | executable_name.resize(executable_name.size() - 4); 362 | } 363 | 364 | #else 365 | // Linux & OSX path delimiter 366 | size_t delimiter_pos = executable_name.find_last_of('/'); 367 | executable_name.erase(0,delimiter_pos+1); 368 | #endif 369 | } 370 | 371 | // Loop over all search paths and return the first hit 372 | for (unsigned int i = 0; i < sizeof(searchPath)/sizeof(char *); ++i) 373 | { 374 | std::string path(searchPath[i]); 375 | size_t executable_name_pos = path.find(""); 376 | 377 | // If there is executable_name variable in the searchPath 378 | // replace it with the value 379 | if (executable_name_pos != std::string::npos) 380 | { 381 | if (executable_path != 0) 382 | { 383 | path.replace(executable_name_pos, strlen(""), executable_name); 384 | } 385 | else 386 | { 387 | // Skip this path entry if no executable argument is given 388 | continue; 389 | } 390 | } 391 | 392 | #ifdef _DEBUG 393 | printf("sdkFindFilePath <%s> in %s\n", filename, path.c_str()); 394 | #endif 395 | 396 | // Test if the file exists 397 | path.append(filename); 398 | FILE *fp; 399 | FOPEN(fp, path.c_str(), "rb"); 400 | 401 | if (fp != NULL) 402 | { 403 | fclose(fp); 404 | // File found 405 | // returning an allocated array here for backwards compatibility reasons 406 | char *file_path = (char *) malloc(path.length() + 1); 407 | STRCPY(file_path, path.length() + 1, path.c_str()); 408 | return file_path; 409 | } 410 | 411 | if (fp) 412 | { 413 | fclose(fp); 414 | } 415 | } 416 | 417 | // File not found 418 | return 0; 419 | } 420 | 421 | #endif -------------------------------------------------------------------------------- /adaptive_gridsampler/setup.py: -------------------------------------------------------------------------------- 1 | import os 2 | import torch 3 | 4 | from setuptools import setup 5 | from torch.utils.cpp_extension import BuildExtension, CUDAExtension 6 | 7 | cxx_args = ['-std=c++11'] 8 | 9 | nvcc_args = [ 10 | '-gencode', 'arch=compute_60,code=sm_60', 11 | '-gencode', 'arch=compute_61,code=sm_61', 12 | '-gencode', 'arch=compute_70,code=sm_70', 13 | '-gencode', 'arch=compute_70,code=compute_70' 14 | ] 15 | 16 | setup( 17 | name='adaptive_gridsampler_cuda', 18 | ext_modules=[ 19 | CUDAExtension('adaptive_gridsampler_cuda', ['adaptive_gridsampler_cuda.cpp', 'adaptive_gridsampler_kernel.cu'], extra_compile_args={'cxx': cxx_args, 'nvcc': nvcc_args}) 20 | ], 21 | cmdclass={'build_ext': BuildExtension} 22 | ) 23 | -------------------------------------------------------------------------------- /figs/overview.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunwj/CAR/05b22776b9f690dac94ced8baeb455bb722c0997/figs/overview.png -------------------------------------------------------------------------------- /figs/qualitative.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunwj/CAR/05b22776b9f690dac94ced8baeb455bb722c0997/figs/qualitative.png -------------------------------------------------------------------------------- /modules.py: -------------------------------------------------------------------------------- 1 | import functools 2 | import numpy as np 3 | 4 | import torch 5 | import torch.nn as nn 6 | 7 | 8 | LEAKY_FACTOR = 0.2 9 | MULT_FACTOR = 1 10 | 11 | 12 | # TEST PASSED 13 | class PixelUnShuffle(nn.Module): 14 | """ 15 | Inverse process of pytorch pixel shuffle module 16 | """ 17 | def __init__(self, down_scale): 18 | """ 19 | :param down_scale: int, down scale factor 20 | """ 21 | super(PixelUnShuffle, self).__init__() 22 | 23 | if not isinstance(down_scale, int): 24 | raise ValueError('Down scale factor must be a integer number') 25 | self.down_scale = down_scale 26 | 27 | def forward(self, input): 28 | """ 29 | :param input: tensor of shape (batch size, channels, height, width) 30 | :return: tensor of shape(batch size, channels * down_scale * down_scale, height / down_scale, width / down_scale) 31 | """ 32 | b, c, h, w = input.size() 33 | assert h % self.down_scale == 0 34 | assert w % self.down_scale == 0 35 | 36 | oc = c * self.down_scale ** 2 37 | oh = int(h / self.down_scale) 38 | ow = int(w / self.down_scale) 39 | 40 | output_reshaped = input.reshape(b, c, oh, self.down_scale, ow, self.down_scale) 41 | output = output_reshaped.permute(0, 1, 3, 5, 2, 4).reshape(b, oc, oh, ow) 42 | 43 | return output 44 | 45 | 46 | class DownsampleBlock(nn.Module): 47 | def __init__(self, scale, input_channels, output_channels, ksize=1): 48 | super(DownsampleBlock, self).__init__() 49 | self.downsample = nn.Sequential( 50 | PixelUnShuffle(scale), 51 | nn.Conv2d(input_channels * (scale ** 2), output_channels, kernel_size=ksize, stride=1, padding=ksize//2) 52 | ) 53 | 54 | def forward(self, input): 55 | return self.downsample(input) 56 | 57 | 58 | class UpsampleBlock(nn.Module): 59 | def __init__(self, scale, input_channels, output_channels, ksize=1): 60 | super(UpsampleBlock, self).__init__() 61 | self.upsample = nn.Sequential( 62 | nn.Conv2d(input_channels, output_channels * (scale ** 2), kernel_size=1, stride=1, padding=ksize//2), 63 | nn.PixelShuffle(scale) 64 | ) 65 | 66 | def forward(self, input): 67 | return self.upsample(input) 68 | 69 | 70 | class ResidualBlock(nn.Module): 71 | def __init__(self, input_channels, channels, ksize=3, 72 | use_instance_norm=False, affine=False): 73 | super(ResidualBlock, self).__init__() 74 | self.channels = channels 75 | self.ksize = ksize 76 | padding = self.ksize // 2 77 | if use_instance_norm: 78 | self.transform = nn.Sequential( 79 | nn.ReflectionPad2d(padding), 80 | nn.Conv2d(input_channels, channels, kernel_size=self.ksize, stride=1), 81 | nn.InstanceNorm2d(channels, affine=affine), 82 | nn.LeakyReLU(0.2), 83 | nn.ReflectionPad2d(padding), 84 | nn.Conv2d(channels, channels, kernel_size=self.ksize, stride=1), 85 | nn.InstanceNorm2d(channels) 86 | ) 87 | else: 88 | self.transform = nn.Sequential( 89 | nn.ReflectionPad2d(padding), 90 | nn.Conv2d(input_channels, channels, kernel_size=self.ksize, stride=1), 91 | nn.LeakyReLU(0.2), 92 | nn.ReflectionPad2d(padding), 93 | nn.Conv2d(channels, channels, kernel_size=self.ksize, stride=1), 94 | ) 95 | 96 | def forward(self, input): 97 | return input + self.transform(input) * MULT_FACTOR 98 | 99 | 100 | class NormalizeBySum(nn.Module): 101 | def forward(self, x): 102 | return x / torch.sum(x, dim=1, keepdim=True).clamp(min=1e-7) 103 | 104 | 105 | class MeanShift(nn.Conv2d): 106 | def __init__(self, rgb_range, rgb_mean=(0.4488, 0.4371, 0.4040), rgb_std=(1.0, 1.0, 1.0), sign=-1): 107 | super(MeanShift, self).__init__(3, 3, kernel_size=1) 108 | std = torch.Tensor(rgb_std) 109 | self.weight.data = torch.eye(3).view(3, 3, 1, 1) / std.view(3, 1, 1, 1) 110 | self.bias.data = sign * rgb_range * torch.Tensor(rgb_mean) / std 111 | for p in self.parameters(): 112 | p.requires_grad = False 113 | 114 | 115 | class DSN(nn.Module): 116 | def __init__(self, k_size, input_channels=3, scale=4): 117 | super(DSN, self).__init__() 118 | 119 | self.k_size = k_size 120 | 121 | self.sub_mean = MeanShift(1) 122 | 123 | self.ds_1 = nn.Sequential( 124 | nn.ReflectionPad2d(2), 125 | nn.Conv2d(input_channels, 64, 5), 126 | nn.LeakyReLU(LEAKY_FACTOR) 127 | ) 128 | 129 | self.ds_2 = DownsampleBlock(2, 64, 128, ksize=1) 130 | self.ds_4 = DownsampleBlock(2, 128, 128, ksize=1) 131 | 132 | res_4 = list() 133 | for idx in range(5): 134 | res_4 += [ResidualBlock(128, 128)] 135 | self.res_4 = nn.Sequential(*res_4) 136 | 137 | self.ds_8 = DownsampleBlock(2, 128, 256) 138 | 139 | self.kernels_trunk = nn.Sequential( 140 | nn.ReflectionPad2d(1), 141 | nn.Conv2d(256, 256, 3), 142 | nn.ReLU(), 143 | nn.ReflectionPad2d(1), 144 | nn.Conv2d(256, 256, 3), 145 | nn.ReLU(), 146 | nn.ReflectionPad2d(1), 147 | nn.Conv2d(256, 256, 3), 148 | nn.ReLU(), 149 | UpsampleBlock(8 // scale, 256, 256, ksize=1), 150 | nn.ReflectionPad2d(1), 151 | nn.Conv2d(256, 256, 3), 152 | nn.ReLU() 153 | ) 154 | 155 | self.kernels_weight = nn.Sequential( 156 | nn.ReflectionPad2d(1), 157 | nn.Conv2d(256, 256, 3), 158 | nn.ReLU(), 159 | nn.ReflectionPad2d(1), 160 | nn.Conv2d(256, k_size ** 2, 3) 161 | ) 162 | 163 | self.offsets_trunk = nn.Sequential( 164 | nn.ReflectionPad2d(1), 165 | nn.Conv2d(256, 256, 3), 166 | nn.ReLU(), 167 | nn.ReflectionPad2d(1), 168 | nn.Conv2d(256, 256, 3), 169 | nn.ReLU(), 170 | nn.ReflectionPad2d(1), 171 | nn.Conv2d(256, 256, 3), 172 | nn.ReLU(), 173 | UpsampleBlock(8 // scale, 256, 256, ksize=1), 174 | nn.ReflectionPad2d(1), 175 | nn.Conv2d(256, 256, 3), 176 | nn.ReLU() 177 | ) 178 | 179 | self.offsets_h_generation = nn.Sequential( 180 | nn.ReflectionPad2d(1), 181 | nn.Conv2d(256, 256, 3), 182 | nn.ReLU(), 183 | nn.ReflectionPad2d(1), 184 | nn.Conv2d(256, k_size ** 2, 3), 185 | nn.Tanh() 186 | ) 187 | 188 | self.offsets_v_generation = nn.Sequential( 189 | nn.ReflectionPad2d(1), 190 | nn.Conv2d(256, 256, 3), 191 | nn.ReLU(), 192 | nn.ReflectionPad2d(1), 193 | nn.Conv2d(256, k_size ** 2, 3), 194 | nn.Tanh() 195 | ) 196 | 197 | def forward(self, x): 198 | x = self.sub_mean(x) 199 | 200 | x = self.ds_1(x) 201 | x = self.ds_2(x) 202 | x = self.ds_4(x) 203 | x = x + self.res_4(x) 204 | x = self.ds_8(x) 205 | 206 | kt = self.kernels_trunk(x) 207 | k_weight = torch.clamp(self.kernels_weight(kt), min=1e-6, max=1) 208 | kernels = k_weight / torch.sum(k_weight, dim=1, keepdim=True).clamp(min=1e-6) 209 | 210 | ot = self.offsets_trunk(x) 211 | offsets_h = self.offsets_h_generation(ot) 212 | offsets_v = self.offsets_v_generation(ot) 213 | 214 | return kernels, offsets_h, offsets_v 215 | -------------------------------------------------------------------------------- /run.py: -------------------------------------------------------------------------------- 1 | import os, argparse 2 | import numpy as np 3 | from tqdm import tqdm 4 | from glob import glob 5 | from PIL import Image 6 | 7 | import torch 8 | import torch.nn as nn 9 | 10 | import utils 11 | from EDSR.edsr import EDSR 12 | from modules import DSN 13 | from adaptive_gridsampler.gridsampler import Downsampler 14 | from skimage.color import rgb2ycbcr 15 | 16 | 17 | parser = argparse.ArgumentParser(description='Content Adaptive Resampler for Image downscaling') 18 | parser.add_argument('--model_dir', type=str, default='./models', help='path to the pre-trained model') 19 | parser.add_argument('--img_dir', type=str, help='path to the HR images to be downscaled') 20 | parser.add_argument('--scale', type=int, help='downscale factor') 21 | parser.add_argument('--output_dir', type=str, help='path to store results') 22 | parser.add_argument('--benchmark', type=bool, default=True, help='report benchmark results') 23 | args = parser.parse_args() 24 | 25 | 26 | SCALE = args.scale 27 | KSIZE = 3 * SCALE + 1 28 | OFFSET_UNIT = SCALE 29 | BENCHMARK = args.benchmark 30 | 31 | kernel_generation_net = DSN(k_size=KSIZE, scale=SCALE).cuda() 32 | downsampler_net = Downsampler(SCALE, KSIZE).cuda() 33 | upscale_net = EDSR(32, 256, scale=SCALE).cuda() 34 | 35 | kernel_generation_net = nn.DataParallel(kernel_generation_net, [0]) 36 | downsampler_net = nn.DataParallel(downsampler_net, [0]) 37 | upscale_net = nn.DataParallel(upscale_net, [0]) 38 | 39 | kernel_generation_net.load_state_dict(torch.load(os.path.join(args.model_dir, '{0}x'.format(SCALE), 'kgn.pth'))) 40 | upscale_net.load_state_dict(torch.load(os.path.join(args.model_dir, '{0}x'.format(SCALE), 'usn.pth'))) 41 | torch.set_grad_enabled(False) 42 | 43 | 44 | def validation(img, name, save_imgs=False, save_dir=None): 45 | kernel_generation_net.eval() 46 | downsampler_net.eval() 47 | upscale_net.eval() 48 | 49 | kernels, offsets_h, offsets_v = kernel_generation_net(img) 50 | downscaled_img = downsampler_net(img, kernels, offsets_h, offsets_v, OFFSET_UNIT) 51 | downscaled_img = torch.clamp(downscaled_img, 0, 1) 52 | downscaled_img = torch.round(downscaled_img * 255) 53 | 54 | reconstructed_img = upscale_net(downscaled_img / 255.0) 55 | 56 | img = img * 255 57 | img = img.data.cpu().numpy().transpose(0, 2, 3, 1) 58 | img = np.uint8(img) 59 | 60 | reconstructed_img = torch.clamp(reconstructed_img, 0, 1) * 255 61 | reconstructed_img = reconstructed_img.data.cpu().numpy().transpose(0, 2, 3, 1) 62 | reconstructed_img = np.uint8(reconstructed_img) 63 | 64 | downscaled_img = downscaled_img.data.cpu().numpy().transpose(0, 2, 3, 1) 65 | downscaled_img = np.uint8(downscaled_img) 66 | 67 | orig_img = img[0, ...].squeeze() 68 | downscaled_img = downscaled_img[0, ...].squeeze() 69 | recon_img = reconstructed_img[0, ...].squeeze() 70 | 71 | if save_imgs and save_dir: 72 | img = Image.fromarray(orig_img) 73 | img.save(os.path.join(save_dir, name + '_orig.png')) 74 | 75 | img = Image.fromarray(downscaled_img) 76 | img.save(os.path.join(save_dir, name + '_down.png')) 77 | 78 | img = Image.fromarray(recon_img) 79 | img.save(os.path.join(save_dir, name + '_recon.png')) 80 | 81 | psnr = utils.cal_psnr(orig_img[SCALE:-SCALE, SCALE:-SCALE, ...], recon_img[SCALE:-SCALE, SCALE:-SCALE, ...], benchmark=BENCHMARK) 82 | 83 | orig_img_y = rgb2ycbcr(orig_img)[:, :, 0] 84 | recon_img_y = rgb2ycbcr(recon_img)[:, :, 0] 85 | orig_img_y = orig_img_y[SCALE:-SCALE, SCALE:-SCALE, ...] 86 | recon_img_y = recon_img_y[SCALE:-SCALE, SCALE:-SCALE, ...] 87 | 88 | ssim = utils.calc_ssim(recon_img_y, orig_img_y) 89 | 90 | return psnr, ssim 91 | 92 | 93 | if __name__ == '__main__': 94 | img_list = glob(os.path.join(args.img_dir, '**', '*.png'), recursive=True) 95 | assert len(img_list) > 0 96 | 97 | if not os.path.exists(args.output_dir): 98 | os.makedirs(args.output_dir) 99 | 100 | psnr_list = list() 101 | ssim_list = list() 102 | for img_file in tqdm(img_list): 103 | name = os.path.basename(img_file) 104 | name = os.path.splitext(name)[0] 105 | 106 | img = utils.load_img(img_file) 107 | 108 | psnr, ssim = validation(img, name, save_imgs=True, save_dir=args.output_dir) 109 | psnr_list.append(psnr) 110 | ssim_list.append(ssim) 111 | 112 | print('Mean PSNR: {0:.2f}'.format(np.mean(psnr_list))) 113 | print('Mean SSIM: {0:.4f}'.format(np.mean(ssim_list))) 114 | -------------------------------------------------------------------------------- /utils.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | import torch 3 | from scipy import signal 4 | from PIL import Image 5 | 6 | 7 | def matlab_style_gauss2D(shape=(3, 3), sigma=0.5): 8 | """ 9 | 2D gaussian mask - should give the same result as MATLAB's fspecial('gaussian',[shape],[sigma]) 10 | Acknowledgement : https://stackoverflow.com/questions/17190649/how-to-obtain-a-gaussian-filter-in-python (Author@ali_m) 11 | """ 12 | m, n = [(ss - 1.) / 2. for ss in shape] 13 | y, x = np.ogrid[-m:m + 1, -n:n + 1] 14 | h = np.exp(-(x * x + y * y) / (2. * sigma * sigma)) 15 | h[h < np.finfo(h.dtype).eps * h.max()] = 0 16 | sumh = h.sum() 17 | if sumh != 0: 18 | h /= sumh 19 | return h 20 | 21 | 22 | def calc_ssim(X, Y, sigma=1.5, K1=0.01, K2=0.03, R=255): 23 | ''' 24 | X : y channel (i.e., luminance) of transformed YCbCr space of X 25 | Y : y channel (i.e., luminance) of transformed YCbCr space of Y 26 | Please follow the setting of psnr_ssim.m in EDSR (Enhanced Deep Residual Networks for Single Image Super-Resolution CVPRW2017). 27 | Official Link : https://github.com/LimBee/NTIRE2017/tree/db34606c2844e89317aac8728a2de562ef1f8aba 28 | The authors of EDSR use MATLAB's ssim as the evaluation tool, 29 | thus this function is the same as ssim.m in MATLAB with C(3) == C(2)/2. 30 | ''' 31 | gaussian_filter = matlab_style_gauss2D((11, 11), sigma) 32 | 33 | X = X.astype(np.float64) 34 | Y = Y.astype(np.float64) 35 | 36 | window = gaussian_filter 37 | 38 | ux = signal.convolve2d(X, window, mode='same', boundary='symm') 39 | uy = signal.convolve2d(Y, window, mode='same', boundary='symm') 40 | 41 | uxx = signal.convolve2d(X * X, window, mode='same', boundary='symm') 42 | uyy = signal.convolve2d(Y * Y, window, mode='same', boundary='symm') 43 | uxy = signal.convolve2d(X * Y, window, mode='same', boundary='symm') 44 | 45 | vx = uxx - ux * ux 46 | vy = uyy - uy * uy 47 | vxy = uxy - ux * uy 48 | 49 | C1 = (K1 * R) ** 2 50 | C2 = (K2 * R) ** 2 51 | 52 | A1, A2, B1, B2 = ((2 * ux * uy + C1, 2 * vxy + C2, ux ** 2 + uy ** 2 + C1, vx + vy + C2)) 53 | D = B1 * B2 54 | S = (A1 * A2) / D 55 | mssim = S.mean() 56 | 57 | return mssim 58 | 59 | 60 | def cal_psnr(img_1, img_2, benchmark=False): 61 | assert img_1.shape[0] == img_2.shape[0] and img_1.shape[1] == img_2.shape[1] 62 | img_1 = np.float64(img_1) 63 | img_2 = np.float64(img_2) 64 | 65 | diff = (img_1 - img_2) / 255.0 66 | if benchmark: 67 | gray_coeff = np.array([65.738, 129.057, 25.064]).reshape(1, 1, 3) / 255.0 68 | diff = diff * gray_coeff 69 | diff = diff[:, :, 0] + diff[:, :, 1] + diff[:, :, 2] 70 | 71 | mse = np.mean(diff ** 2) 72 | psnr = -10.0 * np.log10(mse) 73 | 74 | return psnr 75 | 76 | 77 | def load_img(img_file): 78 | img = Image.open(img_file).convert('RGB') 79 | img = np.array(img) 80 | h, w, _ = img.shape 81 | img = img[:h // 8 * 8, :w // 8 * 8, :] 82 | img = np.array(img) / 255. 83 | img = img.transpose((2, 0, 1)) 84 | img = torch.from_numpy(img).float().unsqueeze(0).cuda() 85 | 86 | return img 87 | --------------------------------------------------------------------------------