{"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/bridging-category-level-and-instance-level","title":"Bridging Category-level and Instance-level Semantic Image Segmentation","arxiv_id":"1605.06885","date":"2016-05-23","proceeding":null,"authors":["Zifeng Wu","Chunhua Shen","Anton Van Den Hengel"],"abstract":"We propose an approach to instance-level image segmentation that is built on\ntop of category-level segmentation. Specifically, for each pixel in a semantic\ncategory mask, its corresponding instance bounding box is predicted using a\ndeep fully convolutional regression network. Thus it follows a different\npipeline to the popular detect-then-segment approaches that first predict\ninstances' bounding boxes, which are the current state-of-the-art in instance\nsegmentation. We show that, by leveraging the strength of our state-of-the-art\nsemantic segmentation models, the proposed method can achieve comparable or\neven better results to detect-then-segment approaches. We make the following\ncontributions. (i) First, we propose a simple yet effective approach to\nsemantic instance segmentation. (ii) Second, we propose an online bootstrapping\nmethod during training, which is critically important for achieving good\nperformance for both semantic category segmentation and instance-level\nsegmentation. (iii) As the performance of semantic category segmentation has a\nsignificant impact on the instance-level segmentation, which is the second step\nof our approach, we train fully convolutional residual networks to achieve the\nbest semantic category segmentation accuracy. On the PASCAL VOC 2012 dataset,\nwe obtain the currently best mean intersection-over-union score of 79.1%. (iv)\nWe also achieve state-of-the-art results for instance-level segmentation.","url_abs":"http://arxiv.org/abs/1605.06885v1","url_pdf":"http://arxiv.org/pdf/1605.06885v1.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":[],"tasks":[{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semantic-segmentation-on-pascal-context","task":"Semantic Segmentation","dataset":"PASCAL Context","model":"VeryDeep","rank_in_archive_order":56,"of":66,"metrics":{"mIoU":"44.5"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1605.06885","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}