{"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/detail-preserving-pooling-in-deep-networks","title":"Detail-Preserving Pooling in Deep Networks","arxiv_id":"1804.04076","date":"2018-04-11","proceeding":"CVPR 2018 6","authors":["Faraz Saeedan","Nicolas Weber","Michael Goesele","Stefan Roth"],"abstract":"Most convolutional neural networks use some method for gradually downscaling\nthe size of the hidden layers. This is commonly referred to as pooling, and is\napplied to reduce the number of parameters, improve invariance to certain\ndistortions, and increase the receptive field size. Since pooling by nature is\na lossy process, it is crucial that each such layer maintains the portion of\nthe activations that is most important for the network's discriminability. Yet,\nsimple maximization or averaging over blocks, max or average pooling, or plain\ndownsampling in the form of strided convolutions are the standard. In this\npaper, we aim to leverage recent results on image downscaling for the purposes\nof deep learning. Inspired by the human visual system, which focuses on local\nspatial changes, we propose detail-preserving pooling (DPP), an adaptive\npooling method that magnifies spatial changes and preserves important\nstructural detail. Importantly, its parameters can be learned jointly with the\nrest of the network. We analyze some of its theoretical properties and show its\nempirical benefits on several datasets and networks, where DPP consistently\noutperforms previous pooling approaches.","url_abs":"http://arxiv.org/abs/1804.04076v1","url_pdf":"http://arxiv.org/pdf/1804.04076v1.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":"detail-preserving-pooling-in-deep-networks","repo_url":"https://github.com/visinf/dpp","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"BSD-3-Clause"}},{"paper_slug":"detail-preserving-pooling-in-deep-networks","repo_url":"https://github.com/gan3sh500/custom-pooling","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1804.04076","atlas_url":"https://app.syntology.ai/?focus=1804.04076","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}