{"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/clockwork-convnets-for-video-semantic","title":"Clockwork Convnets for Video Semantic Segmentation","arxiv_id":"1608.03609","date":"2016-08-11","proceeding":null,"authors":["Evan Shelhamer","Kate Rakelly","Judy Hoffman","Trevor Darrell"],"abstract":"Recent years have seen tremendous progress in still-image segmentation;\nhowever the na\\\"ive application of these state-of-the-art algorithms to every\nvideo frame requires considerable computation and ignores the temporal\ncontinuity inherent in video. We propose a video recognition framework that\nrelies on two key observations: 1) while pixels may change rapidly from frame\nto frame, the semantic content of a scene evolves more slowly, and 2) execution\ncan be viewed as an aspect of architecture, yielding purpose-fit computation\nschedules for networks. We define a novel family of \"clockwork\" convnets driven\nby fixed or adaptive clock signals that schedule the processing of different\nlayers at different update rates according to their semantic stability. We\ndesign a pipeline schedule to reduce latency for real-time recognition and a\nfixed-rate schedule to reduce overall computation. Finally, we extend clockwork\nscheduling to adaptive video processing by incorporating data-driven clocks\nthat can be tuned on unlabeled video. The accuracy and efficiency of clockwork\nconvnets are evaluated on the Youtube-Objects, NYUD, and Cityscapes video\ndatasets.","url_abs":"http://arxiv.org/abs/1608.03609v1","url_pdf":"http://arxiv.org/pdf/1608.03609v1.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":"clockwork-convnets-for-video-semantic","repo_url":"https://github.com/shelhamer/clockwork-fcn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"scheduling","task_name":"Scheduling"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"video-recognition","task_name":"Video Recognition"},{"task_slug":"video-semantic-segmentation","task_name":"Video Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1608.03609","atlas_url":"https://app.syntology.ai/?focus=1608.03609","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}