{"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/an-analysis-of-deep-neural-network-models-for","title":"An Analysis of Deep Neural Network Models for Practical Applications","arxiv_id":"1605.07678","date":"2016-05-24","proceeding":null,"authors":["Alfredo Canziani","Adam Paszke","Eugenio Culurciello"],"abstract":"Since the emergence of Deep Neural Networks (DNNs) as a prominent technique\nin the field of computer vision, the ImageNet classification challenge has\nplayed a major role in advancing the state-of-the-art. While accuracy figures\nhave steadily increased, the resource utilisation of winning models has not\nbeen properly taken into account. In this work, we present a comprehensive\nanalysis of important metrics in practical applications: accuracy, memory\nfootprint, parameters, operations count, inference time and power consumption.\nKey findings are: (1) power consumption is independent of batch size and\narchitecture; (2) accuracy and inference time are in a hyperbolic relationship;\n(3) energy constraint is an upper bound on the maximum achievable accuracy and\nmodel complexity; (4) the number of operations is a reliable estimate of the\ninference time. We believe our analysis provides a compelling set of\ninformation that helps design and engineer efficient DNNs.","url_abs":"http://arxiv.org/abs/1605.07678v4","url_pdf":"http://arxiv.org/pdf/1605.07678v4.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":"an-analysis-of-deep-neural-network-models-for","repo_url":"https://github.com/szagoruyko/imagenet-validation.torch","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok"}},{"paper_slug":"an-analysis-of-deep-neural-network-models-for","repo_url":"https://github.com/CentaurusM/Utils","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"an-analysis-of-deep-neural-network-models-for","repo_url":"https://github.com/aneeqr/Object-Detection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"an-analysis-of-deep-neural-network-models-for","repo_url":"https://github.com/sanjayjonckheere/iN_iS_Tee_One","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1605.07678","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}