{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/unlabeled-samples-generated-by-gan-improve","title":"Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro","arxiv_id":"1701.07717","date":"2017-01-26","proceeding":"ICCV 2017 10","authors":["Zhedong Zheng","Liang Zheng","Yi Yang"],"abstract":"The main contribution of this paper is a simple semi-supervised pipeline that\nonly uses the original training set without collecting extra data. It is\nchallenging in 1) how to obtain more training data only from the training set\nand 2) how to use the newly generated data. In this work, the generative\nadversarial network (GAN) is used to generate unlabeled samples. We propose the\nlabel smoothing regularization for outliers (LSRO). This method assigns a\nuniform label distribution to the unlabeled images, which regularizes the\nsupervised model and improves the baseline. We verify the proposed method on a\npractical problem: person re-identification (re-ID). This task aims to retrieve\na query person from other cameras. We adopt the deep convolutional generative\nadversarial network (DCGAN) for sample generation, and a baseline convolutional\nneural network (CNN) for representation learning. Experiments show that adding\nthe GAN-generated data effectively improves the discriminative ability of\nlearned CNN embeddings. On three large-scale datasets, Market-1501, CUHK03 and\nDukeMTMC-reID, we obtain +4.37%, +1.6% and +2.46% improvement in rank-1\nprecision over the baseline CNN, respectively. We additionally apply the\nproposed method to fine-grained bird recognition and achieve a +0.6%\nimprovement over a strong baseline. The code is available at\nhttps://github.com/layumi/Person-reID_GAN.","url_abs":"http://arxiv.org/abs/1701.07717v5","url_pdf":"http://arxiv.org/pdf/1701.07717v5.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/layumi/Person-reID_GAN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/AI-NERC-NUPT/DDB","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/AI-NERC-NUPT/PFH-OSNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/AI-NERC-NUPT/PLR-OSNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/freeSubmission/SDB","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/hbchen121/dgreid","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/lyy973/OSFA","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"unlabeled-samples-generated-by-gan-improve","repo_url":"https://github.com/whj363636/Adversarial-attack-on-Person-ReID-With-Deep-Mis-Ranking","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"fine-grained-image-classification","task_name":"Fine-Grained Image Classification"},{"task_slug":null,"task_name":"Generative Adversarial Network"},{"task_slug":"person-re-identification","task_name":"Person Re-Identification"},{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[],"datasets_introduced":[{"slug":"dukemtmc-attribute","name":"DukeMTMC-attribute","full_name":"DukeMTMC-attribute"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/fine-grained-image-classification-on-cub-200-1","task":"Fine-Grained Image Classification","dataset":"CUB-200-2011","model":"Basel.+LSRO","rank_in_archive_order":28,"of":30,"metrics":{"Accuracy":"84.4"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-cuhk03","task":"Person Re-Identification","dataset":"CUHK03","model":"VI+LSRO 3","rank_in_archive_order":4,"of":19,"metrics":{"MAP":"87.4","Rank-1":"84.6"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-dukemtmc-reid","task":"Person Re-Identification","dataset":"DukeMTMC-reID","model":"GAN","rank_in_archive_order":86,"of":94,"metrics":{"Rank-1":"67.68","mAP":"47.13"},"uses_additional_data":false},{"leaderboard":"/sota/person-re-identification-on-market-1501","task":"Person Re-Identification","dataset":"Market-1501","model":"GAN","rank_in_archive_order":108,"of":135,"metrics":{"Rank-1":"83.97","mAP":"66.07"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1701.07717","atlas_url":"https://app.syntology.ai/?focus=1701.07717","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}