{"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/hide-and-seek-a-data-augmentation-technique","title":"Hide-and-Seek: A Data Augmentation Technique for Weakly-Supervised Localization and Beyond","arxiv_id":"1811.02545","date":"2018-11-06","proceeding":null,"authors":["Krishna Kumar Singh","Hao Yu","Aron Sarmasi","Gautam Pradeep","Yong Jae Lee"],"abstract":"We propose 'Hide-and-Seek' a general purpose data augmentation technique,\nwhich is complementary to existing data augmentation techniques and is\nbeneficial for various visual recognition tasks. The key idea is to hide\npatches in a training image randomly, in order to force the network to seek\nother relevant content when the most discriminative content is hidden. Our\napproach only needs to modify the input image and can work with any network to\nimprove its performance. During testing, it does not need to hide any patches.\nThe main advantage of Hide-and-Seek over existing data augmentation techniques\nis its ability to improve object localization accuracy in the weakly-supervised\nsetting, and we therefore use this task to motivate the approach. However,\nHide-and-Seek is not tied only to the image localization task, and can\ngeneralize to other forms of visual input like videos, as well as other\nrecognition tasks like image classification, temporal action localization,\nsemantic segmentation, emotion recognition, age/gender estimation, and person\nre-identification. We perform extensive experiments to showcase the advantage\nof Hide-and-Seek on these various visual recognition problems.","url_abs":"http://arxiv.org/abs/1811.02545v1","url_pdf":"http://arxiv.org/pdf/1811.02545v1.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":"hide-and-seek-a-data-augmentation-technique","repo_url":"https://github.com/kkanshul/Hide-and-Seek","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null},{"paper_slug":"hide-and-seek-a-data-augmentation-technique","repo_url":"https://github.com/PaddlePaddle/PaddleClas","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"paddle","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"action-localization","task_name":"Action Localization"},{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"emotion-recognition","task_name":"Emotion Recognition"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"object-localization","task_name":"Object Localization"},{"task_slug":"person-re-identification","task_name":"Person Re-Identification"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1811.02545","atlas_url":"https://app.syntology.ai/?focus=1811.02545","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}