{"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/large-scale-fine-grained-categorization-and","title":"Large Scale Fine-Grained Categorization and Domain-Specific Transfer Learning","arxiv_id":"1806.06193","date":"2018-06-16","proceeding":"CVPR 2018 6","authors":["Yin Cui","Yang song","Chen Sun","Andrew Howard","Serge Belongie"],"abstract":"Transferring the knowledge learned from large scale datasets (e.g., ImageNet)\nvia fine-tuning offers an effective solution for domain-specific fine-grained\nvisual categorization (FGVC) tasks (e.g., recognizing bird species or car make\nand model). In such scenarios, data annotation often calls for specialized\ndomain knowledge and thus is difficult to scale. In this work, we first tackle\na problem in large scale FGVC. Our method won first place in iNaturalist 2017\nlarge scale species classification challenge. Central to the success of our\napproach is a training scheme that uses higher image resolution and deals with\nthe long-tailed distribution of training data. Next, we study transfer learning\nvia fine-tuning from large scale datasets to small scale, domain-specific FGVC\ndatasets. We propose a measure to estimate domain similarity via Earth Mover's\nDistance and demonstrate that transfer learning benefits from pre-training on a\nsource domain that is similar to the target domain by this measure. Our\nproposed transfer learning outperforms ImageNet pre-training and obtains\nstate-of-the-art results on multiple commonly used FGVC datasets.","url_abs":"http://arxiv.org/abs/1806.06193v1","url_pdf":"http://arxiv.org/pdf/1806.06193v1.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":"large-scale-fine-grained-categorization-and","repo_url":"https://github.com/richardaecn/cvpr18-inaturalist-transfer","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"fine-grained-image-classification","task_name":"Fine-Grained Image Classification"},{"task_slug":"fine-grained-visual-categorization","task_name":"Fine-Grained Visual Categorization"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.06193","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}