{"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/unsupervised-video-object-segmentation-for","title":"Unsupervised Video Object Segmentation for Deep Reinforcement Learning","arxiv_id":"1805.07780","date":"2018-05-20","proceeding":"NeurIPS 2018 12","authors":["Vik Goel","Jameson Weng","Pascal Poupart"],"abstract":"We present a new technique for deep reinforcement learning that automatically\ndetects moving objects and uses the relevant information for action selection.\nThe detection of moving objects is done in an unsupervised way by exploiting\nstructure from motion. Instead of directly learning a policy from raw images,\nthe agent first learns to detect and segment moving objects by exploiting flow\ninformation in video sequences. The learned representation is then used to\nfocus the policy of the agent on the moving objects. Over time, the agent\nidentifies which objects are critical for decision making and gradually builds\na policy based on relevant moving objects. This approach, which we call\nMotion-Oriented REinforcement Learning (MOREL), is demonstrated on a suite of\nAtari games where the ability to detect moving objects reduces the amount of\ninteraction needed with the environment to obtain a good policy. Furthermore,\nthe resulting policy is more interpretable than policies that directly map\nimages to actions or values with a black box neural network. We can gain\ninsight into the policy by inspecting the segmentation and motion of each\nobject detected by the agent. This allows practitioners to confirm whether a\npolicy is making decisions based on sensible information.","url_abs":"http://arxiv.org/abs/1805.07780v1","url_pdf":"http://arxiv.org/pdf/1805.07780v1.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":"unsupervised-video-object-segmentation-for","repo_url":"https://github.com/vik-goel/MOREL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"gone","observed_at":"2026-09-18","how":"tree_404+repo_404"}}],"tasks":[{"task_slug":"atari-games","task_name":"Atari Games"},{"task_slug":"decision-making","task_name":"Decision Making"},{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"object","task_name":"Object"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"unsupervised-video-object-segmentation","task_name":"Unsupervised Video Object Segmentation"},{"task_slug":"video-object-segmentation","task_name":"Video Object Segmentation"},{"task_slug":"video-semantic-segmentation","task_name":"Video Semantic Segmentation"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.07780","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}