1<!DOCTYPE html> 2<html> 3 4 <head> 5 <meta charset="utf-8"> 6 <meta http-equiv="X-UA-Compatible" content="IE=edge"> 7 <meta name="viewport" content="width=device-width, initial-scale=1"> 8 <title>Shuguang Dou - Home</title> 9 <meta name="description" content="Shuguang Dou"> 10 <link rel="stylesheet" href="/css/main.css"> 11 <link rel="canonical" href="/"> 12<link rel="icon" type ="image/png" href="/images/favicon.png"> 13 14
15<script>
vendor: 351 bytes, lines 15-21
15 16 (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){ 17 (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), 18 m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m) 19 })(window,document,'script','https://www.google-analytics.com/analytics.js','ga'); 20 21 ga('create', '
21UA-82472331-1
vendor: 42 bytes, lines 21-23
21', 'auto'); 22 ga('send', 'pageview'); 23
24</script>
24 25 26 27</head> 28 29 30 <body> 31 32 <style> 33.navbar-default { 34 background-color: #87CEFA; 35 border-color: #e7e7e7; 36} 37</style> 38<div class="navbar navbar-default navbar-fixed-top" role="navigation"> 39 <div class="container-fluid"> 40 <div class="navbar-header"> 41 <button type="button" class="navbar-toggle collapsed" data-toggle="collapse" data-target="#navbar-collapse-1" aria-expanded="false"> 42 <span class="sr-only">Toggle navigation</span> 43 <span class="icon-bar"></span> 44 <span class="icon-bar"></span> 45 <span class="icon-bar"></span> 46 <span class="icon-bar"></span> 47 </button> 48 49 <a class="navbar-brand" href="/">Shuguang Dou</a> 50 </div> 51 <div class="collapse navbar-collapse" id="navbar-collapse-1"> 52 <ul class="nav navbar-nav navbar-right"> 53 <li><a href="/">Home</a></li> 54 <li><a href="/publications">Publications</a></li> 55 <li><a href="/codes">Codes</a></li> 56 <li><a href="/experience">Experience</a></li> 57 <li><a href="/friends">Friends</a></li> 58 </ul> 59 </div> 60 </div> 61</div> 62 63 64 <div class="container-fluid"> 65 <div class="row"> 66 <div id="gridid" class="col-sm-12"> 67 <div class="container-fluid"> 68 69 <div class="row"> 70 71 <div class="col-sm-8"> 72 <p>I am a Researcher at Algorithm Novelty Lab of Huawei Tech. I obtained my Ph.D. degree from the School of Computer Science and Technology, Tongji University. 73During my PhD studies, I collaborated with researchers <a href="https://scholar.google.com/citations?user=JiTfWVMAAAAJ">Xinyang Jiang</a> and <a href="http://recmind.cn/">Dongsheng Li</a> in theat Microsoft Research Asia (MSRA). 74While interning at MSRA, my research focused on infographic understanding and neural architecture search (NAS) benchmarks. 75I am also a visiting scholar luckily invited by <a href="https://www.cse.msu.edu/~liuxm/index2.html">Prof. Xiaoming Liu</a> of MSU. During my doctoral studies, my research primarily focused on human-centric video understanding, with a specific emphasis on trustworthy person re-identification (ReID). My work addressed the critical pillars of robustness, security, and privacy preservation within this domain. Currently, I am focused on advancing research in the following areas:</p> 76 <ol> 77 <li>Agentic Vision</li> 78 <li>Distillation of Model and Dataset </li> 79 <li>Medical Image Analysis</li> 80</ol> 81 82 <p><strong>We are hiring research scientists, engineers, and research interns for our lab! Please <a href="mailto:[email protected]">drop me a line</a> if you are interested.</strong></p> 83 84 <h3 id="cv">CV</h3> 85 <p>You can download my CV at <a href="https://shuguang-52.github.io/papers/Resume_ShuguangDou.pdf">here</a>.</p> 86 87 <h3 id="news">News</h3> 88 89 <p>May 1, 2026 : 90<em>A paper on âDistilling and Scaling Hierarchical Vision Transformer to 30B Parametersâ is accepted by IJCAI 2026.</em></p> 91 92 <p>Mar 23, 2026 : 93<em>A paper on âDecouple Re-identification and Human Parsing for Occluded Person Re-Identificationâ is accepted by IEEE TBOIM in 2026.</em></p> 94 95 <p>Feb 21, 2026 : 96<em>A paper on âAgentic Video Summarization via Self-Reflecting Multimodal Understandingâ is accepted by CVPR 2026. Congratulations to my intern Miaotian Guo!</em></p> 97 98 <p>Feb 05, 2026 : 99<em>A paper on âActive Dataset Distillation via Dual-Space Informative Matchingâ is accepted by IEEE TIP in 2026.</em></p> 100 101 <p>Jan 26, 2026 : 102<em>A paper on âAsynchronous Matching with Dynamic Sampling for Multimodal Dataset Distillationâ is accepted by ICLR 2026.</em></p> 103 104 <p>Jan 24, 2026 : 105<em>I have been invited to be a Reviewer for the ECCV 2026 Conference.</em></p> 106 107 <p>Jan 12, 2026 : 108<em>I have been invited to be a Reviewer for the ICML 2026 Conference.</em></p> 109 110 <p>Nov 23, 2025 : 111<em>I have been invited to be a Reviewer for the CVPR 2026 Conference.</em></p> 112 113 <p><a href="/allnews.html">see all news</a></p> 114 115 </div> 116 117 <div class="col-sm-4" style="display:table-cell; vertical-align:middle; text-align:center"> 118 119 <ul style="overflow: hidden"> 120 <img src="/images/profiel_pic.jpg" class="img-responsive" width="100%" /> 121 </ul> 122 123 <p><!-- <br clear="all" /> --></p> 124 125 <p><a href="mailto:[email protected]">[email protected]</a> <br /> 126 <a href="https://scholar.google.com.hk/citations?user=n2YT06EAAAAJ&hl=zh-CN">Google Scholar</a><br /> 127 SH021, Shanghai, China.<br /></p> 128 129 </div> 130 131 </div> 132</div> 133 134<div class="col-sm-12"> 135 136 <h3 id="publications">Publications</h3> 137 138 <div class="col-sm-11 clearfix"> 139 <div class="well"> 140 <pubtit>EA-HAS-Bench and Language-Enhanced Shrinkage Search for Energy-aware NAS</pubtit> 141 142 <p><img src="/images/pubpic/25_TPAMI_LESS.png" class="img-responsive" width="200px" style="float: left" /></p> 143 144 <p>Recent studies have started integrating large language models (LLMs) into AutoML frameworks to enhance model search efficie
144ncy and configuration prediction, yet challenges remain in adapting these methods for energy-efficient searches across vast configuration spaces, as they often neglect energy consumption metrics. As a result, we introduce the Language-Enhanced Shrinkage Search (LESS), a plug-and-play method that utilizes the analytical capabilities of LLMs to enhance the energy efficiency of existing hyperparameter optimization techniques.</p> 145 146 <p><em>S<strong>huguang Dou</strong>, Jiale Zhao, Xinyang Jiang, Junyao Gao, Yuge Zhang, Bo Li, Dongsheng Li</em></p> 147 148 <p>IEEE Transactions on Pattern Analysis and Machine Intelligence (<strong>T-PAMI</strong>) in 2025, CCF A</p> 149 150 <p><a href="">pdf</a> 151 / 152 <a href="">project page</a> 153 / 154 <a href="https://ieeexplore.ieee.org/abstract/document/11122892">Website Online</a></p> 155 156 </div> 157 </div> 158 159 <div class="col-sm-11 clearfix"> 160 <div class="well"> 161 <pubtit>Dpl++:Advancing the network performance via image and label perturbations</pubtit> 162 163 <p><img src="/images/pubpic/25_TPAMI_DPL.png" class="img-responsive" width="200px" style="float: left" /></p> 164 165 <p>We propose a novel paradigm that leverages the benefits of image perturbations for rectifying data distributions. Our method, called DPL (Deep Perturbation Learning), introduces new insights into utilizing image perturbations and focuses on improving generalizability on normal samples, rather than resisting adversarial attacks.</p> 166 167 <p><em>Ding Qi, Jian Li, Junyao Gao, <strong>Shuguang Dou</strong>, Ying Tai, Jianlong Hu, Bo Zhao, Yabiao Wang, Chengjie Wang, Cairong Zhao</em></p> 168 169 <p>IEEE Transactions on Pattern Analysis and Machine Intelligence (<strong>T-PAMI</strong>) in 2025, CCF A</p> 170 171 <p><a href="">pdf</a> 172 / 173 <a href="">project page</a> 174 / 175 <a href="https://ieeexplore.ieee.org/abstract/document/11106184/">Website Online</a></p> 176 177 </div> 178 </div> 179 180 <div class="col-sm-11 clearfix"> 181 <div class="well"> 182 <pubtit>MTA-VPS:A Large-scale Benchmark for Video-Based Person Search</pubtit> 183 184 <p><img src="/images/pubpic/25_CVPRW_VPS.png" class="img-responsive" width="200px" style="float: left" /></p> 185 186 <p>We introduce Video-based Person Search (VPS), a task that tracks individuals and their trajectories in raw video scenes, enhancing performance in practical applications. Due to the lack of suitable datasets and privacy concerns, we created the first VPS dataset using virtual environments, featuring diverse continuous video scenes. Paired with an evaluation framework, this dataset assesses both retrieval performance and trajectory quality.</p> 187 188 <p><em>Ding Qi, <strong>Shuguang Dou</strong>, Jian Liu, Huaixuan Cao, Hao Zhang, Dongsheng Jiang, Cairong Zhao</em></p> 189 190 <p>Proceedings of the Computer Vision and Pattern Recognition Conference Workshop in 2025</p> 191 192 <p><a href="">pdf</a> 193 / 194 <a href="https://github.com/Vill-Lab/MTA-VPS">Dataset</a> 195 / 196 <a href="https://openaccess.thecvf.com/content/CVPR2025W/PVUW/html/Qi_MTA-VPS_A_Large-scale_Benchmark_for_Video-Based_Person_Search_CVPRW_2025_paper.html">Website Online</a></p> 197 198 </div> 199 </div> 200 201 <div class="col-sm-11 clearfix"> 202 <div class="well"> 203 <pubtit>Towards Universal Dataset Distillation via Task-Driven Diffusion</pubtit> 204 205 <p><img src="/images/pubpic/25_CVPR_UniDD.png" class="img-responsive" width="200px" style="float: left" /></p> 206 207 <p>We propose a universal dataset distillation framework, named UniDD, a task-driven diffusion model for diverse DD task</p> 208 209 <p><em>Ding Qi, Jian Li, Junyao Gao, <strong>Shuguang Dou</strong>, Ying Tai, Jianlong Hu, Bo Zhao, Yabiao Wang, Chengjie Wang, Cairong Zhao</em></p> 210 211 <p>Proceedings of the Computer Vision and Pattern Recognition Conference (<strong>CVPR Oral</strong>) in 2025</p> 212 213 <p><a href="">pdf</a> 214 / 215 <a href="">project page</a> 216 / 217 <a href="https://openaccess.thecvf.com/content/CVPR2025/html/Qi_Towards_Universal_Dataset_Distillation_via_Task-Driven_Diffusion_CVPR_2025_paper.html">Website Online</a></p> 218 219 </div> 220 </div> 221 222 <div class="col-sm-11 clearfix"> 223 <div class="well"> 224 <pubtit>One for All:A Review of Large Pre-training Models for Re-Identification</pubtit> 225 226 <p><img src="/images/pubpic/25_WWW_MORE.png" class="img-responsive" width="200px" style="float: left" /></p> 227 228 <p>In this paper, we present a comprehensive survey of large pre-training models for ReID, which we refer to as the âOne for Allâ approach. This paradigm leverages large-scale pre-training models, such as self-supervised or language-image pre-training models, as general feature extractors adaptable to multiple ReID tasks.</p> 229 230 <p><em>Yubing Sun, <strong>Shuguang Dou</strong>, Ding Qi, Yanping Li, Jiale Zhao and Cairong Zhao</em></p> 231 232 <p>Web Conference Workshop on Multimedia Object Re-ID:Advancements, Challenges, and Opportunities (MORE 2025)</p> 233 234 <p><a href="">pdf</a> 235 / 236 <a href="">project page</a> 237 / 238 <a href="https:// github.com/ Vill-Lab/ Awesome-Evolving-ReID">Paper List (GitHub)</a> 239 / 240 <a href="https://dl.acm.org/doi/abs/10.1145/3701716.3717652">Website Online</a></p> 241 242 </div> 243 </div> 244 245 <div class="col-sm-11 clearfix"> 246 <div class="well"> 247 <pubtit>Fetch and Forge:Efficient Dataset Condensation for Object Detection</pubtit> 248 249 <p><img src="/images/pubpic/24_NIPS_DCOD.png" class="img-responsive" width="200px" style="float: left" /></p> 250 251 <p>
251Current research on DC mainly focuses on image classification, with less exploration of object detection. As a remedy, we propose DCOD, the first dataset condensation framework for object detection. It initially storing key localization and classification information into model parameters, and then reconstructing synthetic images via model inversion.</p> 252 253 <p><em>Ding Qi, Jian Li, Jinlong Peng, <strong>Shuguang Dou</strong>, Jialin Li, Bo Zhao, Jiangning Zhang, Yabiao Wang, Chengjie Wang, Cairong Zhao</em></p> 254 255 <p>Neural Information Processing Systems (<strong>NeuPIS</strong>) in 2024, CCF A</p> 256 257 <p><a href="">pdf</a> 258 / 259 <a href="">project page</a> 260 / 261 <a href="">code (GitHub)</a> 262 / 263 <a href="">Website Online</a></p> 264 265 </div> 266 </div> 267 268 <div class="col-sm-11 clearfix"> 269 <div class="well"> 270 <pubtit>Re-ID-leak:Membership Inference Attacks Against Person Re-identification</pubtit> 271 272 <p><img src="/images/pubpic/24_IJCV_ReIDLeak.png" class="img-responsive" width="200px" style="float: left" /></p> 273 274 <p>Most existing MI attack methods focus on classification models, while Re-ID follows a distinct paradigm for training and inference. Re-ID is a fine-grained recognition task that involves complex feature embedding, and the model outputs commonly used by existing MI algorithms, such as logits and losses, are inaccessible during inference. We propose a novel MI attack method based on the distribution of inter-sample similarity, which involves sampling a set of anchor images to represent the similarity distribution that is conditioned on a target image.</p> 275 276 <p><em>Junyao Gao, Xinyang Jiang, <strong>Shuguang Dou</strong>, Dongsheng Li, Duoqian Miao, Cairong Zhao</em></p> 277 278 <p>International Journal of Computer Vision (<strong>IJCV</strong>) in 2024, CCF A</p> 279 280 <p><a href="">pdf</a> 281 / 282 <a href="https://jeoyal.github.io/home/">project page</a> 283 / 284 <a href="">code (GitHub)</a> 285 / 286 <a href="https://link.springer.com/article/10.1007/s11263-024-02115-6">Website Online</a></p> 287 288 </div> 289 </div> 290 291 <div class="col-sm-11 clearfix"> 292 <div class="well"> 293 <pubtit>Reviving static charts into live charts</pubtit> 294 295 <p><img src="/images/pubpic/24_TVCG_LiveChart.png" class="img-responsive" width="200px" style="float: left" /></p> 296 297 <p>Data charts are prevalent across various fields due to their efficacy in conveying complex data relationships. However, static charts may sometimes struggle to engage readers and efficiently present intricate information, potentially resulting in limited understanding. We introduce âLive Charts,â a new format of presentation that decomposes complex information within a chart and explains the information pieces sequentially through rich animations and accompanying audio narration.</p> 298 299 <p><em>Lu Ying, Yun Wang, Haotian Li, <strong>Shuguang Dou</strong>, Haidong Zhang, Xinyang Jiang, Huamin Qu, Yingcai Wu</em></p> 300 301 <p>IEEE Transactions on Visualization and Computer Graphics (<strong>T-VCG</strong>) in 2024, CCF A</p> 302 303 <p><a href="">pdf</a> 304 / 305 <a href="https://yinglu.notion.site/Hi-I-am-Lu-Ying-6ada6350f92a4daa8cc75d07f20ec81b">project page</a> 306 / 307 <a href="">code (GitHub)</a> 308 / 309 <a href="https://ieeexplore.ieee.org/abstract/document/10530507">Website Online</a></p> 310 311 </div> 312 </div> 313 314 <div class="col-sm-11 clearfix"> 315 <div class="well"> 316 <pubtit>Hierarchically Recognizing Vector Graphics and A New Chart-based Vector Graphics Dataset</pubtit> 317 318 <p><img src="/images/pubpic/24_TPAMI_Yolat++.png" class="img-responsive" width="200px" style="float: left" /></p> 319 320 <p>The conventional approach to image recognition has been based on raster graphics, which can suffer from aliasing and information loss when scaled up or down. In this paper, we propose a novel approach that leverages the benefits of vector graphics for object localization and classification. Our method, called YOLaT (You Only Look at Text), takes the textual document of vector graphics as input, rather than rendering it into pixels. We propose YOLaT++ to learn Multi-level Abstraction Feature Learning from Primitive Shapes to Curves and Points. On the other hand, given few public datasets focus on vector graphics, data-driven learning cannot exert its full power on this format. We provide a large-scale and challenging dataset for Chart-based Vector Graphics Detection and Chart Understanding, termed VG-DCU, with vector graphics, raster graphics, annotations, and raw data drawn for creating these vector charts.</p> 321 322 <p><em>S<strong>huguang Dou</strong>, Xinyang Jiang, Lu Liu, Lu Ying, Caihua Shan, Yifei Shen, Xuanyi Dong, Yun Wang, Dongsheng Li, Cairong Zhao</em></p> 323 324 <p>IEEE Transactions on Pattern Analysis and Machine Intelligence (<strong>T-PAMI</strong>) in 2024, CCF A</p> 325 326 <p><a href="">pdf</a> 327 / 328 <a href="http://shuguang-52.github.io/vgdcu">project page</a> 329 / 330 <a href="https://github.com/microsoft/YOLaT-VectorGraphicsRecognition">code (GitHub)</a> 331 / 332 <a href="https://ieeexplore.ieee.org/abstract/document/10508965">Website Online</a></p> 333 334 </div> 335 </div> 336 337 <div class="col-sm-11 clearfix"> 338 <div class="well"> 339 <pubtit>Adaptive Discriminative Regularization for Visual Classification</pubtit> 340 341 <p><img src="/images/pubpic/24_IJCV_ADR.png" class="img-responsive" width="200px" style="float: left" /></p> 342 343 <p>How to improve discriminative feature learning is central in classification. In this paper, we embrace the real-world data distribution setting in that some classes share semantic overlaps due to their similar appearances or concepts. Regarding this hypothesis, we propose a novel regularization to improve discriminative learning. We first calibrate the estimated highest likelihood of one sample based on its semantically neighboring classes, then encourage the overall likelihood predictions to be deterministic by imposing an adaptive exponential penalty.</p> 344 345 <p><em>Qingsong Zhao, Yi Wang, <strong>Shuguang Dou</strong>, Chen Gong, Yin Wang and Cairong Zhao</em></p> 346 347 <p>International Journal of Computer Vision (<strong>IJCV</strong>) in 2024, CCF A</p> 348 349 <p><a href="https://shuguang-52.github.io/papers/24ijcv_adr.pdf">pdf</a> 350 / 351 <a href="https://zqsiat.github.io/">project page</a> 352 / 353 <a href="">code (GitHub)</a> 354 / 355 <a href="https://link.springer.com/article/10.1007/s11263-024-02080-0?utm_source=rct_congratemailt&utm_medium=email&utm_campaign=nonoa_20240513&utm_content=10.1007/s11263-024-02080-0">Website Online</a></p> 356 357 </div> 358 </div> 359 360 <div class="col-sm-11 clearfix"> 361 <div class="well"> 362 <pubtit>Invisible backdoor attack with dynamic triggers against person re-identification</pubtit> 363 364 <p><img src="/images/pubpic/23_TIFS_backdoor.png" class="img-responsive" width="200px" style="float: left" /></p> 365 366 <p>Existing backdoor attack methods follow an all-to-one or all-to-all attack scenario, where all the target classes in the test set have already been seen in the training set. However, ReID is a much more complex fine-grained open-set recognition problem, where the identities in the test set are not contained in the training set. To ameliorate this issue, we propose a novel backdoor attack on deep ReID under a new all-to-unknown scenario, called Dynamic Triggers Invisible Backdoor Attack (DT-IBA). Instead of learning fixed triggers for the target classes from the training set, DT-IBA can dynamically generate new triggers for any unknown identities.</p> 367 368 <p><em>Wenli Sun, Xinyang Jiang, <strong>Shuguang Dou</strong>, Dongsheng Li, Duoqian Miao, Cheng Deng, Cairong Zhao</em></p> 369 370 <p>
370IEEE Transactions on Information Forensics and Security (<strong>T-IFS</strong>) in 2023, CCF A</p> 371 372 <p><a href="https://shuguang-52.github.io/papers/23tifs_backdoor.pdf">pdf</a> 373 / 374 <a href="http://shuguang-52.github.io/backdoor">project page</a> 375 / 376 <a href="https://github.com/Vill-Lab/2023-TIFS-DTIBA">code (GitHub)</a> 377 / 378 <a href="https://ieeexplore.ieee.org/abstract/document/10285514/">Website Online</a></p> 379 380 </div> 381 </div> 382 383 <div class="col-sm-11 clearfix"> 384 <div class="well"> 385 <pubtit>EA-HAS-Bench:Energy-Aware Hyperparameter and Architecture Search Benchmark</pubtit> 386 387 <p><img src="/images/pubpic/23_ICLR_BSC.png" class="img-responsive" width="200px" style="float: left" /></p> 388 389 <p>We present the first large-scale energy-aware benchmark that allows studying AutoML methods to achieve better trade-offs between performance and search energy consumption, named EA-HAS-Bench. EA-HAS-Bench provides a large-scale architecture/hyperparameter joint search space, covering diversified configurations related to energy consumption.</p> 390 391 <p><em>S<strong>huguang Dou</strong>, Xinyang Jiang, Cairong Zhao, Dongsheng Li</em></p> 392 393 <p>International Conference on Learning Representations (<strong>ICLR</strong>) in 2023, <strong>Spotlight</strong></p> 394 395 <p><a href="https://shuguang-52.github.io/papers/23iclr_ea_has_bench.pdf">pdf</a> 396 / 397 <a href="http://shuguang-52.github.io/eahasbench">project page</a> 398 / 399 <a href="https://github.com/microsoft/EA-HAS-Bench">code (GitHub)</a> 400 / 401 <a href="https://openreview.net/forum?id=n-bvaLSCC78">Website Online</a></p> 402 403 </div> 404 </div> 405 406 <div class="col-sm-11 clearfix"> 407 <div class="well"> 408 <pubtit>Human Co-Parsing Guided Alignment for Occluded Person Re-identification</pubtit> 409 410 <p><img src="/images/pubpic/22_tip_hcga.png" class="img-responsive" width="200px" style="float: left" /></p> 411 412 <p>We propose a novel Human Co-parsing Guided Alignment (HCGA) framework that alternately trains the human co-parsing network and the ReID network, where the human co-paring network is trained in a <em>weakly supervised manner</em> to obtain paring results without any extra annotation.</p> 413 414 <p><em>S<strong>huguang Dou</strong>, Cairong Zhao, Xinyang Jiang, Shanshan Zhang, Member, IEEE, Wei-Shi Zheng, Wangmeng Zuo, Senior Member, IEEE</em></p> 415 416 <p>IEEE Transactions on Image Processing (<strong>T-IP</strong>) in 2022, CCF A</p> 417 418 <p><a href="https://shuguang-52.github.io/papers/22tip_hcga.pdf">pdf</a> 419 / 420 <a href="http://shuguang-52.github.io/hcga">project page</a> 421 / 422 <a href="https://github.com/Vill-Lab/2022-TIP-HCGA">code (GitHub)</a> 423 / 424 <a href="https://ieeexplore.ieee.org/abstract/document/9994734">Website Online</a></p> 425 426 </div> 427 </div> 428 429 <div class="col-sm-11 clearfix"> 430 <div class="well"> 431 <pubtit>Detecting Overlapped Objects in X-ray Security Imagery by a Label-aware Mechanism</pubtit> 432 433 <p><img src="/images/pubpic/22_tifs_clcxray.jpg" class="img-responsive" width="200px" style="float: left" /></p> 434 435 <p>One of the key challenges to the X-ray security checkis to detect the overlapped items in backpacks or suitcases in the X-ray images. Most existing methods improve the robustness of models to the object overlapping problem by enhancing the underlying visual information such as colors and edges. However, this strategy ignores the situations in that the objects have similar visual clues as to the background, and objects overlapping each other. Since the two cases rarely appear in existing datasets, we contribute a novel dataset â Cutters and Liquid Containers X-ray Dataset (CLCXray) to complete the related research.</p> 436 437 <p><em>Cairong Zhao*, Liang Zhu*, <strong>Shuguang Dou</strong>, Weihong Deng, and Liang Wang, Fellow, IEEE (*Co-First Authors)</em></p> 438 439 <p>IEEE Transactions on Information Forensics and Security (<strong>T-IFS</strong>) in 2022, CCF A</p> 440 441 <p><a href="https://shuguang-52.github.io/papers/22tifs_clcxray.pdf">pdf</a> 442 / 443 <a href="http://shuguang-52.github.io/clcxray">
443project page</a> 444 / 445 <a href="https://github.com/Vill-Lab/2022-TIFS-CLCXray">code (GitHub)</a> 446 / 447 <a href="https://ieeexplore.ieee.org/document/9722843">Website Online</a></p> 448 449 </div> 450 </div> 451 452 <div class="col-sm-11 clearfix"> 453 <div class="well"> 454 <pubtit>Incremental Generative Occlusion Adversarial Suppression Network for Person ReID</pubtit> 455 456 <p><img src="/images/pubpic/21_tip_igoas.png" class="img-responsive" width="200px" style="float: left" /></p> 457 458 <p>To address the occlusion problem, we propose a novel Incremental Generative Occlusion Adversarial Suppression (IGOAS) network.</p> 459 460 <p><em>Cairong Zhao*, Xinbi Lv*, <strong>Shuguang Dou*</strong>, Shanshan Zhang, Member, IEEE, Jun Wu, Senior Member, IEEE, and Liang Wang, Fellow, IEEE (*Co-First Authors)</em></p> 461 462 <p>Accepted by IEEE Transactions on Image Processing (<strong>T-IP</strong>) in 2022, CCF A</p> 463 464 <p><a href="https://shuguang-52.github.io/papers/21tip_igoas.pdf">pdf</a> 465 / 466 <a href="http://shuguang-52.github.io/igoas">project page</a> 467 / 468 <a href="https://github.com/Vill-Lab/2021-TIP-IGOAS">code (GitHub)</a> 469 / 470 <a href="https://ieeexplore.ieee.org/abstract/document/9397375">Website Online</a></p> 471 472 </div> 473 </div> 474 475 <div class="col-sm-11 clearfix"> 476 <div class="well"> 477 <pubtit>Alternately Updated SpectralâSpatial Convolution Network for the Classification of Hyperspectral Images</pubtit> 478 479 <p><img src="/images/pubpic/aussc_19_remote_sensing.png" class="img-responsive" width="200px" style="float: left" /></p> 480 481 <p>The connection structure in the convolutional layers of most deep learning-based algorithms used for the classification of hyperspectral images (HSIs) has typically been in the forward direction. In this study, an end-to-e
481nd alternately updated spectralâspatial convolutional network (AUSSC) with a recurrent feedback structure is used to learn refined spectral and spatial features for HSI classification.</p> 482 483 <p><em>Wenju Wang, <strong>Shuguang Dou^</strong>, Sen Wang (^Corresponding Author and First Student Author)</em></p> 484 485 <p>Remote sensing 2019</p> 486 487 <p><a href="https://shuguang-52.github.io/papers/19remotesensing_aussc.pdf">pdf</a> 488 / 489 <a href="http://shuguang-52.github.io/aussc">project page</a> 490 / 491 <a href="https://github.com/shuguang-52/AUSSC">code (GitHub)</a> 492 / 493 <a href="https://www.mdpi.com/2072-4292/11/15/1794">Website Online</a></p> 494 495 </div> 496 </div> 497 498 <div class="col-sm-11 clearfix"> 499 <div class="well"> 500 <pubtit>A fast dense spectral-spatial convolutional network framework for hyperspectral image classification</pubtit> 501 502 <p><img src="/images/pubpic/fdssc_18_remote_sensing.png" class="img-responsive" width="200px" style="float: left" /></p> 503 504 <p>To reduce the training time and improve accuracy, in this paper we propose an end-to-e
504nd fast dense spectralâspatial convolution (FDSSC) framework for HSI classification.</p> 505 506 <p><em>Wenju Wang, <strong>Shuguang Dou^</strong>, Zhongmin Jiang and Liujie Sun (^Corresponding Author and First Student Author)</em></p> 507 508 <p>Remote sensing 2018ï¼<strong>ESI Highly Cited Paper Citation 300+</strong>ï¼</p> 509 510 <p><a href="https://shuguang-52.github.io/papers/18remotesensing_fdssc.pdf">pdf</a> 511 / 512 <a href="http://shuguang-52.github.io/fdssc">project page</a> 513 / 514 <a href="https://github.com/shuguang-52/FDSSC">code (GitHub)</a> 515 / 516 <a href="https://www.mdpi.com/2072-4292/10/7/1068">Website Online</a></p> 517 518 </div> 519 </div> 520 521 <p><br clear="all" /></p> 522 523 <h4 id="see-all-publications"><a href="/publications">see all publications</a></h4> 524 525</div> 526 527<div class="col-sm-12"> 528 529 <h3 id="friends">Friends</h3> 530 531 <p>Jiaqing Zhang (https://icey-zhang.github.io/) : 532<em>My girlfriend!</em></p> 533 534 <p>ââââââââââââââââââââââââââââââ : 535<em>This is a dividing line.</em></p> 536 537 <p>YuanPeng Tu (https://yuanpengtu.github.io/) : 538<em>The most abstract man in ViLL Lab.</em></p> 539 540 <p>Zifan Song (https://matcaviar.github.io/) : 541<em>The boy at Vill Lab gets a ten out of ten.</em></p> 542 543 <p>Shuyang Feng : 544<em>The most thoughtful man in ViLL Lab.</em></p> 545 546 <p>Junyao Gao (https://jeoyal.github.io/home/) : 547<em>A sincere person in Vill Lab.</em></p> 548 549 <p>Yubin Wang (https://thomaswangy.github.io/) : 550<em>The most trustworthy person in Vill lab.</em></p> 551 552 <p>Wenli Sun : 553<em>My makeup artist in Vill lab.</em></p> 554 555 <p><a href="/friends.html">see all friends</a></p> 556 557 <h3 id="theses">Theses</h3> 558 559 <div class="col-sm-11 clearfix"> 560 <div class="well"> 561 <pubtit>Research on Feature Learning and Application of Three-dimensional Point Cloud</pubtit> 562 563 <p><img src="/images/pubpic/3D.png" class="img-responsive" width="200px" style="float: left" /></p> 564 565 <p>Learning deep features from 3D point cloud for classification and retrieval. This work will be done before 12/2019.</p> 566 567 <p><em>Shuguang Dou</em></p> 568 569 <p>Master Thesis, College of Communication and Art Design, University of Shanghai for Science and Technology; December, 2019</p> 570 571 <p><a href="">pdf in Chinese (Temporarily unavailable)</a></p> 572 573 </div> 574 </div> 575 576 <p> Â </p> 577 578</div> 579 580</div> 581 582 </div> 583 </div> 584 585 <div id="footer" class="panel"> 586 <div class="panel-footer"> 587 <div class="container-fluid"> 588 <div class="row"> 589 <div class="col-sm-5"> 590 591 <p>© 2025 Shuguang Dou. Site made with <a href="https://jekyllrb.com">Jekyll</a>.</p> 592 <p> </p><p> 593 594 595 </div> 596 <div class="col-sm-5"> 597 </div> 598 <div class="col-sm-5"> 599 </div> 600 <div class="col-sm-5"> 601 </div> 602 </div> 603 </div> 604 </div> 605</div> 606
607<script src="https://ajax.googleapis.com/ajax/libs/jquery/1.11.3/jquery.min.js"></script>
vendor: 2 bytes, line 607
607
608<script src="/js/bootstrap.min.js"></script>
608 609 610 611 </body> 612 613</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.