{"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/deep-radial-embedding-for-visual-sequence","title":"Deep Radial Embedding for Visual Sequence Learning","arxiv_id":null,"date":"2022-11-11","proceeding":"European Conference on Computer Vision 2022 11","authors":["Yuecong Min","Peiqi Jiao","Yanan Li","Xiaotao Wang","Lei Lei","Xiujuan Chai","Xilin Chen"],"abstract":"Connectionist Temporal Classification (CTC) is a popular\r\nobjective function in sequence recognition, which provides supervision\r\nfor unsegmented sequence data through aligning sequence and its corresponding labeling iteratively. The blank class of CTC plays a crucial\r\nrole in the alignment process and is often considered responsible for the\r\npeaky behavior of CTC. In this study, we propose an objective function\r\nnamed RadialCTC that constrains sequence features on a hypersphere\r\nwhile retaining the iterative alignment mechanism of CTC. The learned\r\nfeatures of each non-blank class are distributed on a radial arc from the\r\ncenter of the blank class, which provides a clear geometric interpretation\r\nand makes the alignment process more efficient. Besides, RadialCTC can\r\ncontrol the peaky behavior by simply modifying the logit of the blank\r\nclass. Experimental results of recognition and localization demonstrate\r\nthe effectiveness of RadialCTC on two sequence recognition applications.","url_abs":"https://www.ecva.net/papers/eccv_2022/papers_ECCV/html/5670_ECCV_2022_paper.php","url_pdf":"https://www.ecva.net/papers/eccv_2022/papers_ECCV/papers/136660234.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":"arc","task_name":"ARC"},{"task_slug":"sign-language-recognition","task_name":"Sign Language Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/sign-language-recognition-on-rwth-phoenix","task":"Sign Language Recognition","dataset":"RWTH-PHOENIX-Weather 2014","model":"RadialCTC","rank_in_archive_order":8,"of":22,"metrics":{"Word Error Rate (WER)":"20.2"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}