{"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/spatially-adaptive-filter-units-for-deep","title":"Spatially-Adaptive Filter Units for Deep Neural Networks","arxiv_id":"1711.11473","date":"2017-11-30","proceeding":"CVPR 2018 6","authors":["Domen Tabernik","Matej Kristan","Aleš Leonardis"],"abstract":"Classical deep convolutional networks increase receptive field size by either\ngradual resolution reduction or application of hand-crafted dilated\nconvolutions to prevent increase in the number of parameters. In this paper we\npropose a novel displaced aggregation unit (DAU) that does not require\nhand-crafting. In contrast to classical filters with units (pixels) placed on a\nfixed regular grid, the displacement of the DAUs are learned, which enables\nfilters to spatially-adapt their receptive field to a given problem. We\nextensively demonstrate the strength of DAUs on a classification and semantic\nsegmentation tasks. Compared to ConvNets with regular filter, ConvNets with\nDAUs achieve comparable performance at faster convergence and up to 3-times\nreduction in parameters. Furthermore, DAUs allow us to study deep networks from\nnovel perspectives. We study spatial distributions of DAU filters and analyze\nthe number of parameters allocated for spatial coverage in a filter.","url_abs":"http://arxiv.org/abs/1711.11473v2","url_pdf":"http://arxiv.org/pdf/1711.11473v2.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":"spatially-adaptive-filter-units-for-deep","repo_url":"https://github.com/skokec/DAU-ConvNet","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"spatially-adaptive-filter-units-for-deep","repo_url":"https://github.com/skokec/DAU-ConvNet-caffe","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"dau-convnet","method_name":"DAU-ConvNet"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.11473","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}