{"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/multi-attention-multi-class-constraint-for","title":"Multi-Attention Multi-Class Constraint for Fine-grained Image Recognition","arxiv_id":"1806.05372","date":"2018-06-14","proceeding":"ECCV 2018 9","authors":["Ming Sun","Yuchen Yuan","Feng Zhou","Errui Ding"],"abstract":"Attention-based learning for fine-grained image recognition remains a\nchallenging task, where most of the existing methods treat each object part in\nisolation, while neglecting the correlations among them. In addition, the\nmulti-stage or multi-scale mechanisms involved make the existing methods less\nefficient and hard to be trained end-to-end. In this paper, we propose a novel\nattention-based convolutional neural network (CNN) which regulates multiple\nobject parts among different input images. Our method first learns multiple\nattention region features of each input image through the one-squeeze\nmulti-excitation (OSME) module, and then apply the multi-attention multi-class\nconstraint (MAMC) in a metric learning framework. For each anchor feature, the\nMAMC functions by pulling same-attention same-class features closer, while\npushing different-attention or different-class features away. Our method can be\neasily trained end-to-end, and is highly efficient which requires only one\ntraining stage. Moreover, we introduce Dogs-in-the-Wild, a comprehensive dog\nspecies dataset that surpasses similar existing datasets by category coverage,\ndata volume and annotation quality. This dataset will be released upon\nacceptance to facilitate the research of fine-grained image recognition.\nExtensive experiments are conducted to show the substantial improvements of our\nmethod on four benchmark datasets.","url_abs":"http://arxiv.org/abs/1806.05372v1","url_pdf":"http://arxiv.org/pdf/1806.05372v1.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":"multi-attention-multi-class-constraint-for","repo_url":"https://github.com/xcnkx/fine_grained_classification","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"fine-grained-image-recognition","task_name":"Fine-Grained Image Recognition"},{"task_slug":"metric-learning","task_name":"Metric Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/fine-grained-image-classification-on-stanford","task":"Fine-Grained Image Classification","dataset":"Stanford Cars","model":"MAMC","rank_in_archive_order":67,"of":83,"metrics":{"Accuracy":"93.0%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.05372","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}