{"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/actor-critic-instance-segmentation","title":"Actor-Critic Instance Segmentation","arxiv_id":"1904.05126","date":"2019-04-10","proceeding":"CVPR 2019 6","authors":["Nikita Araslanov","Constantin Rothkopf","Stefan Roth"],"abstract":"Most approaches to visual scene analysis have emphasised parallel processing\nof the image elements. However, one area in which the sequential nature of\nvision is apparent, is that of segmenting multiple, potentially similar and\npartially occluded objects in a scene. In this work, we revisit the recurrent\nformulation of this challenging problem in the context of reinforcement\nlearning. Motivated by the limitations of the global max-matching assignment of\nthe ground-truth segments to the recurrent states, we develop an actor-critic\napproach in which the actor recurrently predicts one instance mask at a time\nand utilises the gradient from a concurrently trained critic network. We\nformulate the state, action, and the reward such as to let the critic model\nlong-term effects of the current prediction and incorporate this information\ninto the gradient signal. Furthermore, to enable effective exploration in the\ninherently high-dimensional action space of instance masks, we learn a compact\nrepresentation using a conditional variational auto-encoder. We show that our\nactor-critic model consistently provides accuracy benefits over the recurrent\nbaseline on standard instance segmentation benchmarks.","url_abs":"http://arxiv.org/abs/1904.05126v1","url_pdf":"http://arxiv.org/pdf/1904.05126v1.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":"actor-critic-instance-segmentation","repo_url":"https://github.com/visinf/acis","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.05126","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}