G

├── Classification_3DSAMPLE.ipynb ├── README.md ├── config.py ├── figure ├── 3D_SAMPLE │ └── cell32 │ │ ├── Normalized_confusion_matrix.svg │ │ ├── acc_1.svg │ │ ├── model.png │ │ ├── model.svg │ │ ├── model3.png │ │ ├── test_asp.jpg │ │ └── train_asp.jpg ├── ModelNet10 │ └── cell32 │ │ ├── Normalized_confusion_matrix.svg │ │ ├── acc.svg │ │ ├── acc_1.svg │ │ ├── loss.svg │ │ ├── model.png │ │ ├── model.svg │ │ ├── model2.png │ │ ├── model3.png │ │ ├── test_asp.jpg │ │ ├── test_asp.svg │ │ ├── train_asp.jpg │ │ └── train_asp.svg ├── ModelNet40 │ └── cell32 │ │ ├── Normalized_confusion_matrix.svg │ │ ├── acc.svg │ │ ├── acc_1.svg │ │ ├── loss.svg │ │ ├── model.png │ │ ├── model2.png │ │ ├── model3.png │ │ ├── test_asp.jpg │ │ └── train_asp.jpg ├── fig1.png ├── fig2.png ├── fig3.png └── fig4.png ├── get_modelnet.sh ├── make_npy_file.ipynb ├── poster_takitani2.pdf ├── sample.txt └── viz.ipynb /README.md: -------------------------------------------------------------------------------- 1 | # Classification3dmodel 2 | 3 | Architecture

44 |

52 | 53 | ↓　Accuracy　↓ 54 | 55 | | ModelNet10 | ModelNet40 | My dataset(3D_SAMPLE) | 56 | | :---: | :---: | :---: | 57 | | 90.1% | 86.8% | 97.6% | 58 | 59 | 60 | # When this does not work 61 | Please check ./config.py 62 | 'DATASET = Modelnet10 or Modelnet40' 63 | Remove the comment out of this． 64 | 3D_SAMPLE is our own dataset. I can not publish it now. 65 | 66 | # P.S. 67 | Forgive me for my poor English... XD 68 | 69 | [My twitter](https://twitter.com/tacky0612) ←　Please contact me if you have any troubles. 70 | -------------------------------------------------------------------------------- /config.py: -------------------------------------------------------------------------------- 1 | # encoding: utf-8 2 | import os 3 | import keras 4 | from keras.optimizers import Adam 5 | from keras.callbacks import LearningRateScheduler, ModelCheckpoint , TensorBoard 6 | 7 | # BASE_DIR = os.path.dirname(os.path.abspath(__file__)) 8 | 9 | class Config(): 10 | 11 | def make_tensorboard(set_dir_name='TensorBoard'): 12 | log_dir = set_dir_name 13 | if os.path.exists(log_dir) == False: 14 | os.makedirs(log_dir) 15 | tensorboard = TensorBoard(log_dir=log_dir, write_graph=True, ) 16 | return tensorboard 17 | 18 | # DATASET = "ModelNet10" 19 | DATASET = "ModelNet40" 20 | # DATASET = "3D_SAMPLE" 21 | 22 | 23 | N_POINTS = 10000 24 | CELL = 32 25 | BATCH_SIZE = 128 26 | EPOCHS = 40 27 | LEARNIG_RATE = 0.001 28 | 29 | 30 | OPTIMIZER = Adam(lr=LEARNIG_RATE) 31 | 32 | 33 | if DATASET == "ModelNet10": 34 | DATA_DIR = "ModelNet10/" 35 | NUM_CLASSES = 10 36 | EXTENSION = ".off" 37 | CLASS_NAME = [ 38 | 'bathtub', 39 | 'chair', 40 | 'dresser', 41 | 'night_stand', 42 | 'table', 43 | 'bed', 44 | 'desk', 45 | 'monitor', 46 | 'sofa', 47 | 'toilet', 48 | ] 49 | 50 | if DATASET == "ModelNet40": 51 | DATA_DIR = "ModelNet40/" 52 | NUM_CLASSES = 40 53 | EXTENSION = ".off" 54 | CLASS_NAME = [ 55 | 'airplane', 56 | 'bookshelf', 57 | 'chair', 58 | 'desk', 59 | 'glass_box', 60 | 'laptop', 61 | 'person', 62 | 'range_hood', 63 | 'stool', 64 | 'tv_stand', 65 | 'bathtub', 66 | 'bottle', 67 | 'cone', 68 | 'door', 69 | 'guitar', 70 | 'mantel', 71 | 'piano', 72 | 'sink', 73 | 'table', 74 | 'vase', 75 | 'bed', 76 | 'bowl', 77 | 'cup', 78 | 'dresser', 79 | 'keyboard', 80 | 'monitor', 81 | 'plant', 82 | 'sofa', 83 | 'tent', 84 | 'wardrobe', 85 | 'bench', 86 | 'car', 87 | 'curtain', 88 | 'flower_pot', 89 | 'lamp', 90 | 'night_stand', 91 | 'radio', 92 | 'stairs', 93 | 'toilet', 94 | 'xbox', 95 | ] 96 | 97 | if DATASET == "3D_SAMPLE": 98 | DATA_DIR = "3D_SAMPLE/" 99 | NUM_CLASSES = 34 100 | EXTENSION = ".stl" 101 | CLASS_NAME = [ 102 | 'auto_valve1', 103 | 'auto_valve2', 104 | 'auto_valve3', 105 | 'bearing', 106 | 'bevel_gear', 107 | 'block', 108 | 'bracket1', 109 | 'bracket2', 110 | 'bushing', 111 | 'cylinder', 112 | 'etc', 113 | 'frame', 114 | 'gear', 115 | 'handle_valve1', 116 | 'handle_valve2', 117 | 'handle_valve3', 118 | 'hinge1', 119 | 'hinge2', 120 | 'nut', 121 | 'pipe', 122 | 'plate', 123 | 'pully1', 124 | 'pully2', 125 | 'robot1', 126 | 'robot2', 127 | 'robot3', 128 | 'robot4', 129 | 'rod', 130 | 'shaft1', 131 | 'shaft2', 132 | 'spring', 133 | 'trigeminal_valve', 134 | 'valve_connector', 135 | 'washer' 136 | ] 137 | 138 | 139 | VOX_DIR = "np_vox/" + DATASET + "/cell" + str(CELL) + "/" 140 | FIG_DIR = "figure/"+ DATASET + "/cell" + str(CELL) + "/" 141 | WEIGHTS_DIR = "weights/" + DATASET + "/cell" + str(CELL) + "/" 142 | MODEL_DIR = "model/"+ DATASET + "/cell" + str(CELL) + "/" 143 | DIST_DIR = "dist/"+ DATASET + "/cell" + str(CELL) + "/" 144 | 145 | # early_stopping 146 | early_stopping = keras.callbacks.EarlyStopping(monitor='val_loss', patience=0, verbose=0, mode='auto') 147 | 148 | #save best model 149 | checkpoint = ModelCheckpoint(filepath = os.path.join(WEIGHTS_DIR, "model-{epoch:02d}.h5"), save_best_only=True) 150 | 151 | # TensorBoard 152 | CALLBACKS = [make_tensorboard(),checkpoint] 153 | 154 | 155 | 156 | 157 | config = Config() 158 | 159 | 160 | -------------------------------------------------------------------------------- /figure/3D_SAMPLE/cell32/model.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/3D_SAMPLE/cell32/model.png -------------------------------------------------------------------------------- /figure/3D_SAMPLE/cell32/model.svg: -------------------------------------------------------------------------------- 1 | 2 | 4 | 6 | 7 | 279 | -------------------------------------------------------------------------------- /figure/3D_SAMPLE/cell32/model3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/3D_SAMPLE/cell32/model3.png -------------------------------------------------------------------------------- /figure/3D_SAMPLE/cell32/test_asp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/3D_SAMPLE/cell32/test_asp.jpg -------------------------------------------------------------------------------- /figure/3D_SAMPLE/cell32/train_asp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/3D_SAMPLE/cell32/train_asp.jpg -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/acc.svg: -------------------------------------------------------------------------------- 1 | 2 | 4 | 5 | 1166 | -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/loss.svg: -------------------------------------------------------------------------------- 1 | 2 | 4 | 5 | 1047 | -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/model.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet10/cell32/model.png -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/model.svg: -------------------------------------------------------------------------------- 1 | 2 | 4 | 6 | 7 | 279 | -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/model2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet10/cell32/model2.png -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/model3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet10/cell32/model3.png -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/test_asp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet10/cell32/test_asp.jpg -------------------------------------------------------------------------------- /figure/ModelNet10/cell32/train_asp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet10/cell32/train_asp.jpg -------------------------------------------------------------------------------- /figure/ModelNet40/cell32/loss.svg: -------------------------------------------------------------------------------- 1 | 2 | 4 | 5 | 1106 | -------------------------------------------------------------------------------- /figure/ModelNet40/cell32/model.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet40/cell32/model.png -------------------------------------------------------------------------------- /figure/ModelNet40/cell32/model2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet40/cell32/model2.png -------------------------------------------------------------------------------- /figure/ModelNet40/cell32/model3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet40/cell32/model3.png -------------------------------------------------------------------------------- /figure/ModelNet40/cell32/test_asp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet40/cell32/test_asp.jpg -------------------------------------------------------------------------------- /figure/ModelNet40/cell32/train_asp.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/ModelNet40/cell32/train_asp.jpg -------------------------------------------------------------------------------- /figure/fig1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/fig1.png -------------------------------------------------------------------------------- /figure/fig2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/fig2.png -------------------------------------------------------------------------------- /figure/fig3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/fig3.png -------------------------------------------------------------------------------- /figure/fig4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/figure/fig4.png -------------------------------------------------------------------------------- /get_modelnet.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | 4 | # download zip files 5 | wget http://3dvision.princeton.edu/projects/2014/3DShapeNets/ModelNet10.zip 6 | wget http://modelnet.cs.princeton.edu/ModelNet40.zip 7 | 8 | # extract zip files 9 | unzip ModelNet10.zip 10 | unzip ModelNet40.zip 11 | -------------------------------------------------------------------------------- /make_npy_file.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Head" 8 | ] 9 | }, 10 | { 11 | "cell_type": "markdown", 12 | "metadata": {}, 13 | "source": [ 14 | "## imports" 15 | ] 16 | }, 17 | { 18 | "cell_type": "code", 19 | "execution_count": 1, 20 | "metadata": {}, 21 | "outputs": [], 22 | "source": [ 23 | "# -*- coding: utf-8 -*-" 24 | ] 25 | }, 26 | { 27 | "cell_type": "code", 28 | "execution_count": 2, 29 | "metadata": {}, 30 | "outputs": [], 31 | "source": [ 32 | "\n", 33 | "import os\n", 34 | "import sys\n", 35 | "import numpy as np\n", 36 | "import glob\n", 37 | "import math\n", 38 | "import random\n", 39 | "import shutil\n", 40 | "import keras\n", 41 | "\n", 42 | "from config import config" 43 | ] 44 | }, 45 | { 46 | "cell_type": "markdown", 47 | "metadata": {}, 48 | "source": [ 49 | "## Tuning" 50 | ] 51 | }, 52 | { 53 | "cell_type": "code", 54 | "execution_count": 3, 55 | "metadata": {}, 56 | "outputs": [], 57 | "source": [ 58 | "#Tuning\n", 59 | "dataset = config.DATASET\n", 60 | "n_points = config.N_POINTS\n", 61 | "cell = config.CELL\n", 62 | "num_classes = config.NUM_CLASSES\n", 63 | "extension = config.EXTENSION\n", 64 | "class_name = config.CLASS_NAME" 65 | ] 66 | }, 67 | { 68 | "cell_type": "code", 69 | "execution_count": 4, 70 | "metadata": {}, 71 | "outputs": [], 72 | "source": [ 73 | "train_or_test = [\"train\",\"test\"]" 74 | ] 75 | }, 76 | { 77 | "cell_type": "code", 78 | "execution_count": 5, 79 | "metadata": {}, 80 | "outputs": [], 81 | "source": [ 82 | "#読み込み，書き込み用ディレクトリ\n", 83 | "data_dir = config.DATA_DIR\n", 84 | "vox_dir = config.VOX_DIR\n", 85 | "fig_dir = config.FIG_DIR\n", 86 | "dist_dir = config.DIST_DIR" 87 | ] 88 | }, 89 | { 90 | "cell_type": "code", 91 | "execution_count": 6, 92 | "metadata": {}, 93 | "outputs": [], 94 | "source": [ 95 | "dir_list = [data_dir , vox_dir , fig_dir , dist_dir]\n", 96 | "for directory in dir_list:\n", 97 | " if os.path.exists(directory) == False:\n", 98 | " os.makedirs(directory)" 99 | ] 100 | }, 101 | { 102 | "cell_type": "code", 103 | "execution_count": 7, 104 | "metadata": {}, 105 | "outputs": [], 106 | "source": [ 107 | "# TrainとTestのデータdirectory作成\n", 108 | "for t in train_or_test:\n", 109 | " for cl in class_name:\n", 110 | " if os.path.exists(data_dir + cl + \"/\" + t + \"/\") == False:\n", 111 | " os.makedirs(data_dir + cl + \"/\" + t + \"/\")" 112 | ] 113 | }, 114 | { 115 | "cell_type": "markdown", 116 | "metadata": { 117 | "heading_collapsed": true 118 | }, 119 | "source": [ 120 | "# Function" 121 | ] 122 | }, 123 | { 124 | "cell_type": "code", 125 | "execution_count": 8, 126 | "metadata": { 127 | "hidden": true 128 | }, 129 | "outputs": [], 130 | "source": [ 131 | "def load_off(filename):\n", 132 | " # read OFF file\n", 133 | " with open(filename,\"r\") as handle:\n", 134 | " off = handle.read().rstrip().split(\"\\n\")\n", 135 | " \n", 136 | " #OFFファイルが不正かどうか判定\n", 137 | " if off[0] != \"OFF\":\n", 138 | "# print(\"{} is broken!!\".format(filename))\n", 139 | " params = list(off[0].split(\" \"))\n", 140 | " n_vertices = int(params[0].strip(\"OFF\"))\n", 141 | " n_faces = int(params[1])\n", 142 | " off.insert(0, \"OFF\")\n", 143 | " \n", 144 | " else:\n", 145 | " #get params and faces\n", 146 | " params = list(map(int, off[1].split(\" \")))\n", 147 | " n_vertices = params[0]\n", 148 | " n_faces = params[1]\n", 149 | "\n", 150 | " # read Vertex coordinates\n", 151 | " vertices = []\n", 152 | " for n in range(n_vertices):\n", 153 | " coords = list(map(float, off[2+n].split()))\n", 154 | " vertices.append(coords)\n", 155 | "\n", 156 | " # read information of faces\n", 157 | " faces = []\n", 158 | " for n in range(n_faces):\n", 159 | " connects = list(map(int, off[2 + n_vertices + n].split(\" \")[1:4]))\n", 160 | " faces.append(connects)\n", 161 | "\n", 162 | " return vertices, faces" 163 | ] 164 | }, 165 | { 166 | "cell_type": "code", 167 | "execution_count": 9, 168 | "metadata": { 169 | "hidden": true 170 | }, 171 | "outputs": [], 172 | "source": [ 173 | "def load_stl(filename):\n", 174 | " # read STL file\n", 175 | " with open(filename,\"r\") as handle:\n", 176 | " stl = handle.read().rstrip().split(\"\\n\")\n", 177 | " \n", 178 | " #get vertice\n", 179 | " vertice = []\n", 180 | " for i in range(len(stl)):\n", 181 | " pool = stl[i].split()\n", 182 | " if pool[0] == \"vertex\":\n", 183 | " vertex = list(map(float, pool[1:]))\n", 184 | " vertice.append(vertex)\n", 185 | " \n", 186 | " seen = []\n", 187 | " unique_vertice = [x for x in vertice if x not in seen and not seen.append(x)]\n", 188 | " \n", 189 | " #get faces\n", 190 | " faces = []\n", 191 | " for n in range(len(stl)):\n", 192 | " if stl[n].split() == ['outer', 'loop']:\n", 193 | " indexes = []\n", 194 | " for i in range(3):\n", 195 | " index = get_index_2d_list(unique_vertice, list(map(float,stl[n+i+1].split()[1:])))\n", 196 | " indexes.append(index)\n", 197 | " faces.append(indexes)\n", 198 | "\n", 199 | " return unique_vertice, faces" 200 | ] 201 | }, 202 | { 203 | "cell_type": "code", 204 | "execution_count": 10, 205 | "metadata": { 206 | "hidden": true 207 | }, 208 | "outputs": [], 209 | "source": [ 210 | "def generate_points(points, face_ids, num_generated_point=1000):\n", 211 | " \n", 212 | " points = np.array(points)\n", 213 | " face_ids = np.array(face_ids)\n", 214 | " \n", 215 | " # compute area\n", 216 | " vec1 = points[face_ids[:,0]] - points[face_ids[:,1]]\n", 217 | " vec2 = points[face_ids[:,1]] - points[face_ids[:,2]]\n", 218 | " cross_product = np.cross(vec1, vec2)\n", 219 | " area = np.sqrt(np.power(cross_product, 2).sum(axis=1))\n", 220 | "\n", 221 | " # cumsum area\n", 222 | " cum_area = np.cumsum(area)\n", 223 | "\n", 224 | " # generate random \n", 225 | " random = np.random.rand(num_generated_point) * cum_area[-1]\n", 226 | "\n", 227 | " # convert random to index\n", 228 | " #random_idx = [bisect.bisect_left(cum_area, x) for x in random]\n", 229 | " random_idx = np.random.choice(np.arange(len(face_ids)), size=num_generated_point, p=area/cum_area[-1])\n", 230 | "\n", 231 | " #import pdb; pdb.set_trace()\n", 232 | "\n", 233 | " # generate points\n", 234 | " r1 = np.tile(np.random.rand(num_generated_point), (3, 1)).T\n", 235 | " r2 = np.tile(np.random.rand(num_generated_point), (3, 1)).T\n", 236 | "\n", 237 | " generated_points = (1 - np.sqrt(r1))*points[face_ids[random_idx, 0]] + np.sqrt(r1)*(1-r2)*points[face_ids[random_idx, 1]] + np.sqrt(r1)*r2*points[face_ids[random_idx, 2]]\n", 238 | "\n", 239 | " return generated_points, random_idx" 240 | ] 241 | }, 242 | { 243 | "cell_type": "code", 244 | "execution_count": 11, 245 | "metadata": { 246 | "hidden": true 247 | }, 248 | "outputs": [], 249 | "source": [ 250 | "def voxilize(np_pc,cell):\n", 251 | "# ボクセル化した配列を返す\n", 252 | " max_dist = 0.0\n", 253 | " for it in range(0,3):\n", 254 | " # 最大値と最小値の距離を求める\n", 255 | " min_ = np.amin(np_pc[:,it])\n", 256 | " max_ = np.amax(np_pc[:,it])\n", 257 | " dist = max_-min_\n", 258 | "\n", 259 | " #xyzで一番並行距離が大きいのを求める\n", 260 | " if dist > max_dist:\n", 261 | " max_dist = dist\n", 262 | " \n", 263 | " for it in range(0,3):\n", 264 | "\n", 265 | " # 最大値と最小値の距離を求める\n", 266 | " min_ = np.amin(np_pc[:,it])\n", 267 | " max_ = np.amax(np_pc[:,it])\n", 268 | " dist = max_-min_\n", 269 | " \n", 270 | " #中心座標を 0,0,0にセットする（原点が中心にくるようにする）\n", 271 | " np_pc[:,it] = np_pc[:,it] - dist/2 - min_\n", 272 | "\n", 273 | " #covered cell\n", 274 | " cls = cell - 3\n", 275 | "\n", 276 | " #ボクセル一個当たりのサイズを求める\n", 277 | " vox_sz = max_dist/(cls-1)\n", 278 | "\n", 279 | " #上で算出した値で各点を割る。これで各点は(-14, 14)の範囲の値になる\n", 280 | " np_pc[:,it] = np_pc[:,it]/vox_sz\n", 281 | "\n", 282 | " #各点が全て正の整数になるよう移動。これで各点は[0, 30]になる（多分）\n", 283 | " np_pc[:,it] = np_pc[:,it] + (cls-1)/2\n", 284 | "\n", 285 | "\n", 286 | " #整数にする\n", 287 | " np_pc = np.rint(np_pc).astype(np.uint32)\n", 288 | "\n", 289 | "\n", 290 | " #３０＊３０＊３０の配列を作り，点が存在する場合は1、存在しない場合は0を入力する。\n", 291 | " vox = np.zeros([cell-2,cell-2,cell-2])\n", 292 | "\n", 293 | " # (pc_x, pc_y, pc_z)にnp_pcの座標を代入する\n", 294 | " for (pc_x, pc_y, pc_z) in np_pc:\n", 295 | "\n", 296 | "# # 点が存在しても20%の確率で0とし、データにノイズを加え、汎用性を上げている\n", 297 | "# # ここ，ノイジーなデータの場合８０よりも小さい数字にしたほうがいいかもね？\n", 298 | "# if random.randint(0,100) < 80:\n", 299 | " vox[pc_x, pc_y, pc_z] = 1\n", 300 | "\n", 301 | " np_vox = np.zeros([1,cell,cell,cell,1])\n", 302 | " np_vox[0, 1:-1, 1:-1, 1:-1,0] = vox\n", 303 | "\n", 304 | " return np_vox" 305 | ] 306 | }, 307 | { 308 | "cell_type": "code", 309 | "execution_count": 12, 310 | "metadata": { 311 | "hidden": true 312 | }, 313 | "outputs": [], 314 | "source": [ 315 | "def voxel_scatter(np_vox):\n", 316 | "# キレイに整形するやつ\n", 317 | " #空の配列を作る\n", 318 | " vox_scat = np.zeros([0,3], dtype= np.uint32)\n", 319 | "\n", 320 | " #32回\n", 321 | " for x in range(0,np_vox.shape[1]):\n", 322 | " #32回\n", 323 | " for y in range(0,np_vox.shape[2]):\n", 324 | " #32回\n", 325 | " for z in range(0,np_vox.shape[3]):\n", 326 | " #（ｘ，ｙ，ｚ）に１が入っていればその座標を返す\n", 327 | " if np_vox[0,x,y,z,0] == 1.0:\n", 328 | " arr_tmp = np.zeros([1,3],dtype=np.uint32)\n", 329 | " arr_tmp[0,:] = (x,y,z)\n", 330 | " vox_scat = np.concatenate((vox_scat,arr_tmp))\n", 331 | " return vox_scat" 332 | ] 333 | }, 334 | { 335 | "cell_type": "code", 336 | "execution_count": 13, 337 | "metadata": { 338 | "hidden": true 339 | }, 340 | "outputs": [], 341 | "source": [ 342 | "def load_vox_stl(filename, n_points = n_points):\n", 343 | " #shape(N,1,32,32,32)の配列を返す\n", 344 | " vertices, faces = load_stl(filename)\n", 345 | " points,faces = generate_points(vertices, faces, n_points)\n", 346 | " # ValueError: sequence too large; cannot be greater than 32の回避策\n", 347 | " # list 2 numpy.ndarray\n", 348 | " pc = np.empty((len(points), len(points[0])))\n", 349 | " pc[:] = points\n", 350 | " vox = voxilize(pc,cell)\n", 351 | "\n", 352 | " return vox" 353 | ] 354 | }, 355 | { 356 | "cell_type": "code", 357 | "execution_count": 14, 358 | "metadata": { 359 | "hidden": true 360 | }, 361 | "outputs": [], 362 | "source": [ 363 | "def load_vox_off(filename, n_points = n_points):\n", 364 | " #shape(N,1,32,32,32)の配列を返す\n", 365 | " vertices, faces = load_off(filename)\n", 366 | " points,faces = generate_points(vertices, faces, n_points)\n", 367 | " # ValueError: sequence too large; cannot be greater than 32の回避策\n", 368 | " # list 2 numpy.ndarray\n", 369 | " pc = np.empty((len(points), len(points[0])))\n", 370 | " pc[:] = points\n", 371 | " vox = voxilize(pc,cell)\n", 372 | "\n", 373 | " return vox" 374 | ] 375 | }, 376 | { 377 | "cell_type": "code", 378 | "execution_count": 15, 379 | "metadata": { 380 | "hidden": true 381 | }, 382 | "outputs": [], 383 | "source": [ 384 | "# XYZ 長さの取得\n", 385 | "def calc_dist_off(filename):\n", 386 | " vertices, faces = load_off(filename)\n", 387 | " point = np.array(vertices)\n", 388 | " dist = np.zeros(3)\n", 389 | " for i in range(0,3):\n", 390 | " min_ = np.amin(point[:,i])\n", 391 | " max_ = np.amax(point[:,i])\n", 392 | " dist_ = max_ - min_\n", 393 | " if dist_ < 0.001:\n", 394 | " dist_ = 0.001\n", 395 | " dist[i] = dist_\n", 396 | " return dist" 397 | ] 398 | }, 399 | { 400 | "cell_type": "code", 401 | "execution_count": 16, 402 | "metadata": {}, 403 | "outputs": [], 404 | "source": [ 405 | "# XYZ 長さの取得\n", 406 | "def calc_dist_stl(filename):\n", 407 | " vertices, faces = load_stl(filename)\n", 408 | " point = np.array(vertices)\n", 409 | " dist = np.zeros(3)\n", 410 | " for i in range(0,3):\n", 411 | " min_ = np.amin(point[:,i])\n", 412 | " max_ = np.amax(point[:,i])\n", 413 | " dist_ = max_ - min_\n", 414 | " if dist_ < 0.001:\n", 415 | " dist_ = 0.001\n", 416 | " dist[i] = dist_\n", 417 | " return dist" 418 | ] 419 | }, 420 | { 421 | "cell_type": "markdown", 422 | "metadata": {}, 423 | "source": [ 424 | "# ディレクトリ操作" 425 | ] 426 | }, 427 | { 428 | "cell_type": "code", 429 | "execution_count": null, 430 | "metadata": {}, 431 | "outputs": [], 432 | "source": [ 433 | "\n", 434 | " " 435 | ] 436 | }, 437 | { 438 | "cell_type": "code", 439 | "execution_count": 17, 440 | "metadata": { 441 | "scrolled": true 442 | }, 443 | "outputs": [], 444 | "source": [ 445 | "\n", 446 | "if dataset == \"3D_SAMPLE\":\n", 447 | " # train_or_testの分ける\n", 448 | " for cl in class_name:\n", 449 | " namelist = glob.glob(data_dir+ cl +\"/*\")\n", 450 | " namelist.remove(data_dir+ cl + \"/train\")\n", 451 | " namelist.remove(data_dir+ cl + \"/test\")\n", 452 | " for i in range(len(namelist)):\n", 453 | " if i%5 == 0 :\n", 454 | " shutil.move(namelist[i], data_dir + cl + \"/test/.\")\n", 455 | " # print(namelist[i])\n", 456 | " else:\n", 457 | " shutil.move(namelist[i], data_dir + cl + \"/train/.\")\n", 458 | " # print(namelist[i])\n", 459 | "# nameの取得\n", 460 | "namelist_all = []\n", 461 | "\n", 462 | "for cl in class_name:\n", 463 | " \n", 464 | " for t in train_or_test:\n", 465 | " i = 0\n", 466 | " #ファイル名を取得\n", 467 | " namelist_class = glob.glob(data_dir+ cl +\"/\" + t + \"/*\")\n", 468 | " namelist_class.sort()\n", 469 | " namelist_all.append(namelist_class)\n", 470 | "\n" 471 | ] 472 | }, 473 | { 474 | "cell_type": "markdown", 475 | "metadata": {}, 476 | "source": [ 477 | "# ボクセルのnumpy保存" 478 | ] 479 | }, 480 | { 481 | "cell_type": "code", 482 | "execution_count": 23, 483 | "metadata": {}, 484 | "outputs": [ 485 | { 486 | "name": "stdout", 487 | "output_type": "stream", 488 | "text": [ 489 | "airplane\n", 490 | "bookshelf\n", 491 | "chair\n", 492 | "desk\n", 493 | "glass_box\n", 494 | "laptop\n", 495 | "person\n", 496 | "range_hood\n", 497 | "stool\n", 498 | "tv_stand\n", 499 | "bathtub\n", 500 | "bottle\n", 501 | "cone\n", 502 | "door\n", 503 | "guitar\n", 504 | "mantel\n", 505 | "piano\n", 506 | "sink\n", 507 | "table\n", 508 | "vase\n", 509 | "bed\n", 510 | "bowl\n", 511 | "cup\n", 512 | "dresser\n", 513 | "keyboard\n", 514 | "monitor\n", 515 | "plant\n", 516 | "sofa\n", 517 | "tent\n", 518 | "wardrobe\n", 519 | "bench\n", 520 | "car\n", 521 | "curtain\n", 522 | "flower_pot\n", 523 | "lamp\n", 524 | "night_stand\n", 525 | "radio\n", 526 | "stairs\n", 527 | "toilet\n", 528 | "xbox\n", 529 | "airplane\n", 530 | "bookshelf\n", 531 | "chair\n", 532 | "desk\n", 533 | "glass_box\n", 534 | "laptop\n", 535 | "person\n", 536 | "range_hood\n", 537 | "stool\n", 538 | "tv_stand\n", 539 | "bathtub\n", 540 | "bottle\n", 541 | "cone\n", 542 | "door\n", 543 | "guitar\n", 544 | "mantel\n", 545 | "piano\n", 546 | "sink\n", 547 | "table\n", 548 | "vase\n", 549 | "bed\n", 550 | "bowl\n", 551 | "cup\n", 552 | "dresser\n", 553 | "keyboard\n", 554 | "monitor\n", 555 | "plant\n", 556 | "sofa\n", 557 | "tent\n", 558 | "wardrobe\n", 559 | "bench\n", 560 | "car\n", 561 | "curtain\n", 562 | "flower_pot\n", 563 | "lamp\n", 564 | "night_stand\n", 565 | "radio\n", 566 | "stairs\n", 567 | "toilet\n", 568 | "xbox\n" 569 | ] 570 | } 571 | ], 572 | "source": [ 573 | "off_voxilize = False\n", 574 | "\n", 575 | "for t in train_or_test:\n", 576 | " for cl in class_name:\n", 577 | " print(cl)\n", 578 | " \n", 579 | " #すでにボクセル化が済んていればスキップ\n", 580 | " if os.path.exists(vox_dir + \"x_train.npy\") == True:\n", 581 | "# print(\"{} is exist.\".format(vox_dir + cl + t + \".npy\"))\n", 582 | " off_voxilize = True\n", 583 | " continue\n", 584 | " \n", 585 | " #すでに存在してるnpyファイルはスキップ\n", 586 | " if os.path.exists(vox_dir + cl + t + \".npy\") == True:\n", 587 | " print(\"{} is exist.\".format(data_dir + cl + t + \".npy\"))\n", 588 | " continue\n", 589 | " \n", 590 | " data_3d = glob.glob(data_dir+ cl +\"/\" + t + \"/*\")\n", 591 | " data_3d.sort()\n", 592 | " flag0 = True\n", 593 | " \n", 594 | " for d in data_3d:\n", 595 | " if flag0 :\n", 596 | " #stl,off判定\n", 597 | " if extension == \".off\":\n", 598 | " vox_data = load_vox_off(d)\n", 599 | " elif extension == \".stl\":\n", 600 | " vox_data = load_vox_stl(d)\n", 601 | " flag0 = False\n", 602 | " continue\n", 603 | " if extension == \".off\":\n", 604 | " vox_data_ = load_vox_off(d)\n", 605 | " elif extension == \".stl\":\n", 606 | " vox_data_ = load_vox_stl(d) \n", 607 | " \n", 608 | " vox_data = np.append(vox_data, vox_data_, axis=0)\n", 609 | " np.save(vox_dir + cl + t + \".npy\", vox_data)\n", 610 | "#クラス毎，TestTrain毎ボクセル化されたnpzが生成される（N,32,32,32,1）" 611 | ] 612 | }, 613 | { 614 | "cell_type": "code", 615 | "execution_count": 19, 616 | "metadata": {}, 617 | "outputs": [ 618 | { 619 | "data": { 620 | "text/plain": [ 621 | "'.off'" 622 | ] 623 | }, 624 | "execution_count": 19, 625 | "metadata": {}, 626 | "output_type": "execute_result" 627 | } 628 | ], 629 | "source": [ 630 | "extension" 631 | ] 632 | }, 633 | { 634 | "cell_type": "code", 635 | "execution_count": 22, 636 | "metadata": {}, 637 | "outputs": [ 638 | { 639 | "data": { 640 | "text/plain": [ 641 | "[]" 642 | ] 643 | }, 644 | "execution_count": 22, 645 | "metadata": {}, 646 | "output_type": "execute_result" 647 | } 648 | ], 649 | "source": [ 650 | "data_3d" 651 | ] 652 | }, 653 | { 654 | "cell_type": "code", 655 | "execution_count": null, 656 | "metadata": {}, 657 | "outputs": [], 658 | "source": [] 659 | }, 660 | { 661 | "cell_type": "code", 662 | "execution_count": 37, 663 | "metadata": {}, 664 | "outputs": [ 665 | { 666 | "ename": "FileNotFoundError", 667 | "evalue": "[Errno 2] No such file or directory: 'np_vox/ModelNet40/cell32/airplanetrain.npy'", 668 | "output_type": "error", 669 | "traceback": [ 670 | "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", 671 | "\u001b[0;31mFileNotFoundError\u001b[0m Traceback (most recent call last)", 672 | "\u001b[0;32m\u001b[0m in \u001b[0;36m\u001b[0;34m\u001b[0m\n\u001b[1;32m 8\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mcl\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mclass_name\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 9\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mswich_npy\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m \u001b[0mnpy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvox_dir\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mcl\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mt\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;34m\".npy\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 11\u001b[0m \u001b[0mnum_of_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnpy\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 12\u001b[0m \u001b[0mswich_npy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", 673 | "\u001b[0;32m~/anaconda3/lib/python3.6/site-packages/numpy/lib/npyio.py\u001b[0m in \u001b[0;36mload\u001b[0;34m(file, mmap_mode, allow_pickle, fix_imports, encoding)\u001b[0m\n\u001b[1;32m 370\u001b[0m \u001b[0mown_fid\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 371\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfile\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbasestring\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 372\u001b[0;31m \u001b[0mfid\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfile\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"rb\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 373\u001b[0m \u001b[0mown_fid\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 374\u001b[0m \u001b[0;32melif\u001b[0m \u001b[0mis_pathlib_path\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfile\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", 674 | "\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: 'np_vox/ModelNet40/cell32/airplanetrain.npy'" 675 | ] 676 | } 677 | ], 678 | "source": [ 679 | "\n", 680 | "#ココらへん可読性わるすぎ\n", 681 | "if not off_voxilize:\n", 682 | " # データ整形\n", 683 | " for t in train_or_test:\n", 684 | " swich_npy = True \n", 685 | " num_of_data = []#init\n", 686 | " for cl in class_name:\n", 687 | " if swich_npy == True:\n", 688 | " npy = np.load(vox_dir + cl + t + \".npy\")\n", 689 | " num_of_data.append(npy.shape[0])\n", 690 | " swich_npy = False\n", 691 | " else:\n", 692 | " npy_ = np.load(vox_dir + cl + t + \".npy\")\n", 693 | " npy = np.append(npy,npy_,axis=0)\n", 694 | " num_of_data.append(npy_.shape[0])\n", 695 | " if t == \"train\":\n", 696 | " x_train = npy\n", 697 | " y_train = num_of_data\n", 698 | " if t == \"test\":\n", 699 | " x_test = npy\n", 700 | " y_test = num_of_data\n", 701 | " \n", 702 | " \n", 703 | " # ラベルの生成\n", 704 | " for t in train_or_test:\n", 705 | " if t == \"train\":\n", 706 | " for i in range(num_classes):\n", 707 | " if i == 0:\n", 708 | " label = np.full(y_train[i] , i ) \n", 709 | " else:\n", 710 | " label_ = np.full(y_train[i] , i ) \n", 711 | " label = np.append(label, label_, axis=0)\n", 712 | " y_train = label\n", 713 | "\n", 714 | " if t == \"test\":\n", 715 | " for i in range(num_classes):\n", 716 | " if i == 0:\n", 717 | " label = np.full(y_test[i] , i ) \n", 718 | " else:\n", 719 | " label_ = np.full(y_test[i] , i ) \n", 720 | " label = np.append(label, label_, axis=0)\n", 721 | " y_test = label\n", 722 | "\n", 723 | "\n", 724 | "# convert class vectors to binary class matrices\n", 725 | " y_train = keras.utils.to_categorical(y_train, num_classes)\n", 726 | " y_test = keras.utils.to_categorical(y_test, num_classes)\n", 727 | " \n", 728 | " # 保存\n", 729 | " npy = [\"x_train.npy\" , \"x_test.npy\" , \"y_train.npy\" , \"y_test.npy\"]\n", 730 | " data = [x_train , x_test , y_train , y_test]\n", 731 | " for i in range(len(npy)):\n", 732 | " np.save(vox_dir + npy[i],data[i])\n", 733 | " \n", 734 | " # 不要ファイルの除去\n", 735 | " for i in os.listdir(vox_dir):\n", 736 | " if not i in npy:\n", 737 | " os.remove(vox_dir + i)" 738 | ] 739 | }, 740 | { 741 | "cell_type": "markdown", 742 | "metadata": {}, 743 | "source": [ 744 | "# 距離の計算" 745 | ] 746 | }, 747 | { 748 | "cell_type": "code", 749 | "execution_count": null, 750 | "metadata": {}, 751 | "outputs": [], 752 | "source": [ 753 | "dist_calcrated = False\n", 754 | "\n", 755 | "for t in train_or_test:\n", 756 | " for cl in class_name:\n", 757 | " print(cl)\n", 758 | " \n", 759 | " #すでに距離の計算が済んていればスキップ\n", 760 | " if os.path.exists(dist_dir + \"x_train.npy\") == True:\n", 761 | "# print(\"{} is exist.\".format(vox_dir + cl + t + \".npy\"))\n", 762 | " dist_calcrated = True\n", 763 | " continue\n", 764 | " \n", 765 | " #すでに存在してるnpyファイルはスキップ\n", 766 | " if os.path.exists(dist_dir + cl + t + \".npy\") == True:\n", 767 | "# print(\"{} is exist.\".format(data_dir + cl + t + \".npy\"))\n", 768 | " continue\n", 769 | " \n", 770 | " data_3d = glob.glob(data_dir+ cl +\"/\" + t + \"/*\")\n", 771 | " data_3d.sort()\n", 772 | " flag0 = True\n", 773 | " for d in data_3d:\n", 774 | " if flag0:\n", 775 | " if extension == \".off\":\n", 776 | " dist = calc_dist_off(d)\n", 777 | " elif extension == \".stl\":\n", 778 | " dist = calc_dist_stl(d)\n", 779 | " dist = [dist]\n", 780 | " flag0 = False\n", 781 | " continue\n", 782 | " if extension == \".off\":\n", 783 | " dist_ = calc_dist_off(d)\n", 784 | " elif extension == \".stl\":\n", 785 | " dist_ = calc_dist_stl(d)\n", 786 | " \n", 787 | " dist = np.append(dist, [dist_], axis=0)\n", 788 | "# print(dist.shape)\n", 789 | " np.save(dist_dir + cl + t + \".npy\", dist)" 790 | ] 791 | }, 792 | { 793 | "cell_type": "code", 794 | "execution_count": 19, 795 | "metadata": {}, 796 | "outputs": [ 797 | { 798 | "ename": "NameError", 799 | "evalue": "name 'dist_calcrated' is not defined", 800 | "output_type": "error", 801 | "traceback": [ 802 | "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", 803 | "\u001b[0;31mNameError\u001b[0m Traceback (most recent call last)", 804 | "\u001b[0;32m\u001b[0m in \u001b[0;36m\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mdist_calcrated\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 2\u001b[0m \u001b[0;31m# データ整形\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 3\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mt\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtrain_or_test\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 4\u001b[0m \u001b[0mswich_npy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 5\u001b[0m \u001b[0mnum_of_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;31m#init\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", 805 | "\u001b[0;31mNameError\u001b[0m: name 'dist_calcrated' is not defined" 806 | ] 807 | } 808 | ], 809 | "source": [ 810 | "if not dist_calcrated:\n", 811 | " # データ整形\n", 812 | " for t in train_or_test:\n", 813 | " swich_npy = True \n", 814 | " num_of_data = []#init\n", 815 | " for cl in class_name:\n", 816 | " if swich_npy == True:\n", 817 | " npy = np.load(dist_dir + cl + t + \".npy\")\n", 818 | " num_of_data.append(npy.shape[0])\n", 819 | " swich_npy = False\n", 820 | " else:\n", 821 | " npy_ = np.load(dist_dir + cl + t + \".npy\")\n", 822 | " npy = np.append(npy,npy_,axis=0)\n", 823 | " num_of_data.append(npy_.shape[0])\n", 824 | " if t == \"train\":\n", 825 | " x_train = npy\n", 826 | " y_train = num_of_data\n", 827 | " if t == \"test\":\n", 828 | " x_test = npy\n", 829 | " y_test = num_of_data\n", 830 | " \n", 831 | " \n", 832 | " # ラベルの生成\n", 833 | " for t in train_or_test:\n", 834 | " if t == \"train\":\n", 835 | " for i in range(num_classes):\n", 836 | " if i == 0:\n", 837 | " label = np.full(y_train[i] , i ) \n", 838 | " else:\n", 839 | " label_ = np.full(y_train[i] , i ) \n", 840 | " label = np.append(label, label_, axis=0)\n", 841 | " y_train = label\n", 842 | "\n", 843 | " if t == \"test\":\n", 844 | " for i in range(num_classes):\n", 845 | " if i == 0:\n", 846 | " label = np.full(y_test[i] , i ) \n", 847 | " else:\n", 848 | " label_ = np.full(y_test[i] , i ) \n", 849 | " label = np.append(label, label_, axis=0)\n", 850 | " y_test = label\n", 851 | "\n", 852 | "\n", 853 | "# convert class vectors to binary class matrices\n", 854 | " y_train = keras.utils.to_categorical(y_train, num_classes)\n", 855 | " y_test = keras.utils.to_categorical(y_test, num_classes)\n", 856 | " \n", 857 | " # 保存\n", 858 | " npy = [\"x_train.npy\" , \"x_test.npy\" , \"y_train.npy\" , \"y_test.npy\"]\n", 859 | " data = [x_train , x_test , y_train , y_test]\n", 860 | " for i in range(len(npy)):\n", 861 | " np.save(dist_dir + npy[i],data[i])\n", 862 | " \n", 863 | " # 不要ファイルの除去\n", 864 | "# for i in os.listdir(dist_dir):\n", 865 | "# if not i in npy:\n", 866 | "# os.remove(dist_dir + i)" 867 | ] 868 | }, 869 | { 870 | "cell_type": "code", 871 | "execution_count": null, 872 | "metadata": {}, 873 | "outputs": [], 874 | "source": [] 875 | }, 876 | { 877 | "cell_type": "code", 878 | "execution_count": null, 879 | "metadata": {}, 880 | "outputs": [], 881 | "source": [] 882 | }, 883 | { 884 | "cell_type": "code", 885 | "execution_count": null, 886 | "metadata": {}, 887 | "outputs": [], 888 | "source": [] 889 | }, 890 | { 891 | "cell_type": "code", 892 | "execution_count": null, 893 | "metadata": {}, 894 | "outputs": [], 895 | "source": [] 896 | }, 897 | { 898 | "cell_type": "code", 899 | "execution_count": null, 900 | "metadata": {}, 901 | "outputs": [], 902 | "source": [] 903 | }, 904 | { 905 | "cell_type": "code", 906 | "execution_count": null, 907 | "metadata": {}, 908 | "outputs": [], 909 | "source": [] 910 | }, 911 | { 912 | "cell_type": "code", 913 | "execution_count": null, 914 | "metadata": {}, 915 | "outputs": [], 916 | "source": [] 917 | }, 918 | { 919 | "cell_type": "code", 920 | "execution_count": null, 921 | "metadata": {}, 922 | "outputs": [], 923 | "source": [] 924 | }, 925 | { 926 | "cell_type": "code", 927 | "execution_count": null, 928 | "metadata": {}, 929 | "outputs": [], 930 | "source": [] 931 | }, 932 | { 933 | "cell_type": "code", 934 | "execution_count": null, 935 | "metadata": {}, 936 | "outputs": [], 937 | "source": [] 938 | }, 939 | { 940 | "cell_type": "code", 941 | "execution_count": null, 942 | "metadata": {}, 943 | "outputs": [], 944 | "source": [] 945 | }, 946 | { 947 | "cell_type": "code", 948 | "execution_count": null, 949 | "metadata": {}, 950 | "outputs": [], 951 | "source": [] 952 | }, 953 | { 954 | "cell_type": "code", 955 | "execution_count": null, 956 | "metadata": {}, 957 | "outputs": [], 958 | "source": [] 959 | } 960 | ], 961 | "metadata": { 962 | "kernelspec": { 963 | "display_name": "Python 3", 964 | "language": "python", 965 | "name": "python3" 966 | }, 967 | "language_info": { 968 | "codemirror_mode": { 969 | "name": "ipython", 970 | "version": 3 971 | }, 972 | "file_extension": ".py", 973 | "mimetype": "text/x-python", 974 | "name": "python", 975 | "nbconvert_exporter": "python", 976 | "pygments_lexer": "ipython3", 977 | "version": "3.6.8" 978 | }, 979 | "toc": { 980 | "base_numbering": 1, 981 | "nav_menu": {}, 982 | "number_sections": true, 983 | "sideBar": true, 984 | "skip_h1_title": false, 985 | "title_cell": "Table of Contents", 986 | "title_sidebar": "Contents", 987 | "toc_cell": false, 988 | "toc_position": {}, 989 | "toc_section_display": true, 990 | "toc_window_display": false 991 | } 992 | }, 993 | "nbformat": 4, 994 | "nbformat_minor": 2 995 | } 996 | -------------------------------------------------------------------------------- /poster_takitani2.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/tacky0612/classification3dmodel/4e4e4c37d305703a4afdd445a24ac11a81dd9f6c/poster_takitani2.pdf -------------------------------------------------------------------------------- /sample.txt: -------------------------------------------------------------------------------- 1 | hoge. 2 | --------------------------------------------------------------------------------