{"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-metric-learning-with-angular-loss","title":"Deep Metric Learning with Angular Loss","arxiv_id":"1708.01682","date":"2017-08-04","proceeding":"ICCV 2017 10","authors":["Jian Wang","Feng Zhou","Shilei Wen","Xiao Liu","Yuanqing Lin"],"abstract":"The modern image search system requires semantic understanding of image, and\na key yet under-addressed problem is to learn a good metric for measuring the\nsimilarity between images. While deep metric learning has yielded impressive\nperformance gains by extracting high level abstractions from image data, a\nproper objective loss function becomes the central issue to boost the\nperformance. In this paper, we propose a novel angular loss, which takes angle\nrelationship into account, for learning better similarity metric. Whereas\nprevious metric learning methods focus on optimizing the similarity\n(contrastive loss) or relative similarity (triplet loss) of image pairs, our\nproposed method aims at constraining the angle at the negative point of triplet\ntriangles. Several favorable properties are observed when compared with\nconventional methods. First, scale invariance is introduced, improving the\nrobustness of objective against feature variance. Second, a third-order\ngeometric constraint is inherently imposed, capturing additional local\nstructure of triplet triangles than contrastive loss or triplet loss. Third,\nbetter convergence has been demonstrated by experiments on three publicly\navailable datasets.","url_abs":"http://arxiv.org/abs/1708.01682v1","url_pdf":"http://arxiv.org/pdf/1708.01682v1.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":"deep-metric-learning-with-angular-loss","repo_url":"https://github.com/tomp11/metric_learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"image-retrieval","task_name":"Image Retrieval"},{"task_slug":"metric-learning","task_name":"Metric Learning"},{"task_slug":null,"task_name":"Triplet"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1708.01682","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}