{"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/fast-zero-shot-image-tagging","title":"Fast Zero-Shot Image Tagging","arxiv_id":"1605.09759","date":"2016-05-31","proceeding":"CVPR 2016 6","authors":["Yang Zhang","Boqing Gong","Mubarak Shah"],"abstract":"The well-known word analogy experiments show that the recent word vectors\ncapture fine-grained linguistic regularities in words by linear vector offsets,\nbut it is unclear how well the simple vector offsets can encode visual\nregularities over words. We study a particular image-word relevance relation in\nthis paper. Our results show that the word vectors of relevant tags for a given\nimage rank ahead of the irrelevant tags, along a principal direction in the\nword vector space. Inspired by this observation, we propose to solve image\ntagging by estimating the principal direction for an image. Particularly, we\nexploit linear mappings and nonlinear deep neural networks to approximate the\nprincipal direction from an input image. We arrive at a quite versatile tagging\nmodel. It runs fast given a test image, in constant time w.r.t.\\ the training\nset size. It not only gives superior performance for the conventional tagging\ntask on the NUS-WIDE dataset, but also outperforms competitive baselines on\nannotating images with previously unseen tags","url_abs":"http://arxiv.org/abs/1605.09759v1","url_pdf":"http://arxiv.org/pdf/1605.09759v1.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":"multi-label-zero-shot-learning","task_name":"Multi-label zero-shot learning"},{"task_slug":"zero-shot-learning","task_name":"Zero-Shot Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/multi-label-zero-shot-learning-on-nus-wide","task":"Multi-label zero-shot learning","dataset":"NUS-WIDE","model":"fast0tag","rank_in_archive_order":10,"of":10,"metrics":{"mAP":"15.1"},"uses_additional_data":false},{"leaderboard":"/sota/multi-label-zero-shot-learning-on-open-images","task":"Multi-label zero-shot learning","dataset":"Open Images V4","model":"Fast0tag","rank_in_archive_order":5,"of":8,"metrics":{"MAP":"41.2"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1605.09759","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}