{"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/visualbackprop-efficient-visualization-of","title":"VisualBackProp: efficient visualization of CNNs","arxiv_id":"1611.05418","date":"2016-11-16","proceeding":null,"authors":["Mariusz Bojarski","Anna Choromanska","Krzysztof Choromanski","Bernhard Firner","Larry Jackel","Urs Muller","Karol Zieba"],"abstract":"This paper proposes a new method, that we call VisualBackProp, for\nvisualizing which sets of pixels of the input image contribute most to the\npredictions made by the convolutional neural network (CNN). The method heavily\nhinges on exploring the intuition that the feature maps contain less and less\nirrelevant information to the prediction decision when moving deeper into the\nnetwork. The technique we propose was developed as a debugging tool for\nCNN-based systems for steering self-driving cars and is therefore required to\nrun in real-time, i.e. it was designed to require less computations than a\nforward propagation. This makes the presented visualization method a valuable\ndebugging tool which can be easily used during both training and inference. We\nfurthermore justify our approach with theoretical arguments and theoretically\nconfirm that the proposed method identifies sets of input pixels, rather than\nindividual pixels, that collaboratively contribute to the prediction. Our\ntheoretical findings stand in agreement with the experimental results. The\nempirical evaluation shows the plausibility of the proposed approach on the\nroad video data as well as in other applications and reveals that it compares\nfavorably to the layer-wise relevance propagation approach, i.e. it obtains\nsimilar visualization results and simultaneously achieves order of magnitude\nspeed-ups.","url_abs":"http://arxiv.org/abs/1611.05418v3","url_pdf":"http://arxiv.org/pdf/1611.05418v3.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":"visualbackprop-efficient-visualization-of","repo_url":"https://github.com/AlexeyZhuravlev/visual-backprop","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"visualbackprop-efficient-visualization-of","repo_url":"https://github.com/devansh20la/VisualBackprop","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"visualbackprop-efficient-visualization-of","repo_url":"https://github.com/griffinbran/Democratizing-Autonomous-Driving","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"visualbackprop-efficient-visualization-of","repo_url":"https://github.com/griffinbran/machines_best_friend","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"self-driving-cars","task_name":"Self-Driving Cars"}],"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}