{"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/from-volcano-to-toyshop-adaptive","title":"From Volcano to Toyshop: Adaptive Discriminative Region Discovery for Scene Recognition","arxiv_id":"1807.08624","date":"2018-07-23","proceeding":null,"authors":["Zhengyu Zhao","Martha Larson"],"abstract":"As deep learning approaches to scene recognition emerge, they have continued\nto leverage discriminative regions at multiple scales, building on practices\nestablished by conventional image classification research. However, approaches\nremain largely generic, and do not carefully consider the special properties of\nscenes. In this paper, inspired by the intuitive differences between scenes and\nobjects, we propose Adi-Red, an adaptive approach to discriminative region\ndiscovery for scene recognition. Adi-Red uses a CNN classifier, which was\npre-trained using only image-level scene labels, to discover discriminative\nimage regions directly. These regions are then used as a source of features to\nperform scene recognition. The use of the CNN classifier makes it possible to\nadapt the number of discriminative regions per image using a simple, yet\nelegant, threshold, at relatively low computational cost. Experimental results\non the scene recognition benchmark dataset SUN397 demonstrate the ability of\nAdi-Red to outperform the state of the art. Additional experimental analysis on\nthe Places dataset reveals the advantages of Adi-Red, and highlight how they\nare specific to scenes. We attribute the effectiveness of Adi-Red to the\nability of adaptive region discovery to avoid introducing noise, while also not\nmissing out on important information.","url_abs":"http://arxiv.org/abs/1807.08624v2","url_pdf":"http://arxiv.org/pdf/1807.08624v2.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":"from-volcano-to-toyshop-adaptive","repo_url":"https://github.com/ZhengyuZhao/Adi-Red-Scene","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"attribute","task_name":"Attribute"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"scene-recognition","task_name":"Scene Recognition"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}