{"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/weakly-supervised-instance-segmentation-using-1","title":"Weakly Supervised Instance Segmentation using Class Peak Response","arxiv_id":"1804.00880","date":"2018-04-03","proceeding":"CVPR 2018 6","authors":["Yanzhao Zhou","Yi Zhu","Qixiang Ye","Qiang Qiu","Jianbin Jiao"],"abstract":"Weakly supervised instance segmentation with image-level labels, instead of\nexpensive pixel-level masks, remains unexplored. In this paper, we tackle this\nchallenging problem by exploiting class peak responses to enable a\nclassification network for instance mask extraction. With image labels\nsupervision only, CNN classifiers in a fully convolutional manner can produce\nclass response maps, which specify classification confidence at each image\nlocation. We observed that local maximums, i.e., peaks, in a class response map\ntypically correspond to strong visual cues residing inside each instance.\nMotivated by this, we first design a process to stimulate peaks to emerge from\na class response map. The emerged peaks are then back-propagated and\neffectively mapped to highly informative regions of each object instance, such\nas instance boundaries. We refer to the above maps generated from class peak\nresponses as Peak Response Maps (PRMs). PRMs provide a fine-detailed\ninstance-level representation, which allows instance masks to be extracted even\nwith some off-the-shelf methods. To the best of our knowledge, we for the first\ntime report results for the challenging image-level supervised instance\nsegmentation task. Extensive experiments show that our method also boosts\nweakly supervised pointwise localization as well as semantic segmentation\nperformance, and reports state-of-the-art results on popular benchmarks,\nincluding PASCAL VOC 2012 and MS COCO.","url_abs":"http://arxiv.org/abs/1804.00880v1","url_pdf":"http://arxiv.org/pdf/1804.00880v1.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":"weakly-supervised-instance-segmentation-using-1","repo_url":"https://github.com/ZhouYanzhao/PRM","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-level-supervised-instance-segmentation","task_name":"Image-level Supervised Instance Segmentation"},{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"weakly-supervised-instance-segmentation","task_name":"Weakly-supervised instance segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-level-supervised-instance-segmentation","task":"Image-level Supervised Instance Segmentation","dataset":"PASCAL VOC 2012 val","model":"PRM","rank_in_archive_order":13,"of":13,"metrics":{"mAP@0.25":"44.3","mAP@0.5":"26.8","mAP@0.75":"9.0"},"uses_additional_data":true},{"leaderboard":"/sota/semantic-segmentation-on-pascal-voc-2012-val","task":"Semantic Segmentation","dataset":"PASCAL VOC 2012 val","model":"PRM","rank_in_archive_order":27,"of":29,"metrics":{"mIoU":"53.4%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1804.00880","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}