{"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/optimizing-video-object-detection-via-a-scale","title":"Optimizing Video Object Detection via a Scale-Time Lattice","arxiv_id":"1804.05472","date":"2018-04-16","proceeding":"CVPR 2018 6","authors":["Kai Chen","Jiaqi Wang","Shuo Yang","Xingcheng Zhang","Yuanjun Xiong","Chen Change Loy","Dahua Lin"],"abstract":"High-performance object detection relies on expensive convolutional networks\nto compute features, often leading to significant challenges in applications,\ne.g. those that require detecting objects from video streams in real time. The\nkey to this problem is to trade accuracy for efficiency in an effective way,\ni.e. reducing the computing cost while maintaining competitive performance. To\nseek a good balance, previous efforts usually focus on optimizing the model\narchitectures. This paper explores an alternative approach, that is, to\nreallocate the computation over a scale-time space. The basic idea is to\nperform expensive detection sparsely and propagate the results across both\nscales and time with substantially cheaper networks, by exploiting the strong\ncorrelations among them. Specifically, we present a unified framework that\nintegrates detection, temporal propagation, and across-scale refinement on a\nScale-Time Lattice. On this framework, one can explore various strategies to\nbalance performance and cost. Taking advantage of this flexibility, we further\ndevelop an adaptive scheme with the detector invoked on demand and thus obtain\nimproved tradeoff. On ImageNet VID dataset, the proposed method can achieve a\ncompetitive mAP 79.6% at 20 fps, or 79.0% at 62 fps as a performance/speed\ntradeoff.","url_abs":"http://arxiv.org/abs/1804.05472v1","url_pdf":"http://arxiv.org/pdf/1804.05472v1.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":"optimizing-video-object-detection-via-a-scale","repo_url":"https://github.com/guanfuchen/video_obj","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"object","task_name":"Object"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"video-object-detection","task_name":"Video Object Detection"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1804.05472","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}