{"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/return-of-the-devil-in-the-details-delving","title":"Return of the Devil in the Details: Delving Deep into Convolutional Nets","arxiv_id":"1405.3531","date":"2014-05-14","proceeding":null,"authors":["Ken Chatfield","Karen Simonyan","Andrea Vedaldi","Andrew Zisserman"],"abstract":"The latest generation of Convolutional Neural Networks (CNN) have achieved\nimpressive results in challenging benchmarks on image recognition and object\ndetection, significantly raising the interest of the community in these\nmethods. Nevertheless, it is still unclear how different CNN methods compare\nwith each other and with previous state-of-the-art shallow representations such\nas the Bag-of-Visual-Words and the Improved Fisher Vector. This paper conducts\na rigorous evaluation of these new techniques, exploring different deep\narchitectures and comparing them on a common ground, identifying and disclosing\nimportant implementation details. We identify several useful properties of\nCNN-based representations, including the fact that the dimensionality of the\nCNN output layer can be reduced significantly without having an adverse effect\non performance. We also identify aspects of deep and shallow methods that can\nbe successfully shared. In particular, we show that the data augmentation\ntechniques commonly applied to CNN-based methods can also be applied to shallow\nmethods, and result in an analogous performance boost. Source code and models\nto reproduce the experiments in the paper is made publicly available.","url_abs":"http://arxiv.org/abs/1405.3531v4","url_pdf":"http://arxiv.org/pdf/1405.3531v4.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":"return-of-the-devil-in-the-details-delving","repo_url":"https://github.com/tzing/t-cnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1405.3531","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}