113 | TIPS: Text-Induced Pose Synthesis 114 |
115 |
116 | Prasun Roy 1
117 | Subhankar Ghosh 1
118 | Saumik Bhattacharya 2
119 | Umapada Pal 3
120 | Michael Blumenstein 1
121 |
122 | 1 University of Technology Sydney
123 |
124 | 2 Indian Institute of Technology Kharagpur
125 |
126 | 3 Indian Statistical Institute Kolkata
127 |
128 |
129 | The European Conference on Computer Vision (ECCV) 2022
130 |
138 | 144 | Abstract 145 |
146 |147 | In computer vision, human pose synthesis and transfer deal with probabilistic image generation of a person in a previously unseen pose from an already available observation of that person. Though researchers have recently proposed several methods to achieve this task, most of these techniques derive the target pose directly from the desired target image on a specific dataset, making the underlying process challenging to apply in real-world scenarios as the generation of the target image is the actual aim. In this paper, we first present the shortcomings of current pose transfer algorithms and then propose a novel text-based pose transfer technique to address those issues. We divide the problem into three independent stages: (a) text to pose representation, (b) pose refinement, and (c) pose rendering. To the best of our knowledge, this is one of the first attempts to develop a text-based pose transfer framework where we also introduce a new dataset DF-PASS, by adding descriptive pose annotations for the images of the DeepFashion dataset. The proposed method generates promising results with significant qualitative and quantitative scores in our experiments. 148 |
149 |154 | Network Architecture 155 |
156 |158 |
161 |
162 | 164 |
165 | Click on the diagram for a zoomed view of the network architecture (Opens a new tab). 166 |
167 |168 | The workflow is divided into three stages. In stage 1, we estimate a spatial representation \(K^*_B\) for the target pose \(P_B\) from the corresponding text description embedding \(v_B\). In stage 2, we regressively refine the initial estimation of the facial keypoints to obtain the refined target keypoints \(\tilde{K}^*_B\). Finally, in stage 3, we render the target image \(\tilde{I}_B\) by conditioning the pose transfer on the source image \(I_A\) having the keypoints \(K_A\) corresponding to the source pose \(P_A\). 169 |
170 |176 | Generation Results 177 |
178 |180 |
182 | 184 |
185 | Keypoints-guided methods tend to produce structurally inaccurate results when the physical appearance of the target pose reference significantly differs from the condition image. This observation is more frequent for the out of distribution target poses than the within distribution target poses. On the other hand, the existing text-guided method occasionally misinterprets the target pose due to a limited set of basic poses used for pose representation. The proposed text-guided technique successfully addresses these issues while retaining the ability to generate visually decent results close to the keypoints-guided baseline. 186 |
187 |193 | Paper and Supplementary Materials 194 |
195 |196 |
255 | Citation 256 |
257 |
258 | @inproceedings{roy2022tips,
259 | title = {TIPS: Text-Induced Pose Synthesis},
260 | author = {Roy, Prasun and Ghosh, Subhankar and Bhattacharya, Saumik and Pal, Umapada and Blumenstein, Michael},
261 | booktitle = {The European Conference on Computer Vision (ECCV)},
262 | month = {October},
263 | year = {2022}
264 | }
265 |
266 | 271 | Video Presentation 272 |
273 |279 | Poster Presentation 280 |
281 |289 | News and Updates 290 |
291 |294 | Jul 25, 2022 295 |
296 |
297 | We have released our paper, supplementary materials, code, datasets and pretrained models.
298 |
299 | Star
300 | Fork
301 |
305 | Jul 4, 2022 306 |
307 |
308 | Our paper is accepted in ECCV 2022.
309 |
310 | More details about the code and datasets will be released soon.
311 |
317 | On Twitter 318 |
319 |321 |Happy to share our #ECCV2022 work "TIPS: Text-Induced Pose Synthesis"!
— Prasun Roy (@_prasunroy) July 26, 2022
arXiv - https://t.co/bU8KI0i0ED
Project - https://t.co/XM2pQaZEEP
322 |Thrilled 🤩 to have our 1/1 paper accepted in #ECCV2022.
— Prasun Roy (@_prasunroy) July 4, 2022
Congratulations to my amazing teammates Saumik and Subhankar. Many thanks to my advisors Prof Blumenstein and Prof Pal.
More on the paper coming soon!@UTSEngage @UTSFEIT @ProfBlumenstein
328 | Copyright 2022 by the authors | 329 | Made with on Earth. 330 |
331 |