{"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/caa-channelized-axial-attention-for-semantic","title":"Channelized Axial Attention for Semantic Segmentation -- Considering Channel Relation within Spatial Attention for Semantic Segmentation","arxiv_id":"2101.07434","date":"2021-01-19","proceeding":null,"authors":["Ye Huang","Di Kang","Wenjing Jia","Xiangjian He","Liu Liu"],"abstract":"Spatial and channel attentions, modelling the semantic interdependencies in spatial and channel dimensions respectively, have recently been widely used for semantic segmentation. However, computing spatial and channel attentions separately sometimes causes errors, especially for those difficult cases. In this paper, we propose Channelized Axial Attention (CAA) to seamlessly integrate channel attention and spatial attention into a single operation with negligible computation overhead. Specifically, we break down the dot-product operation of the spatial attention into two parts and insert channel relation in between, allowing for independently optimized channel attention on each spatial location. We further develop grouped vectorization, which allows our model to run with very little memory consumption without slowing down the running speed. Comparative experiments conducted on multiple benchmark datasets, including Cityscapes, PASCAL Context, and COCO-Stuff, demonstrate that our CAA outperforms many state-of-the-art segmentation models (including dual attention) on all tested datasets.","url_abs":"https://arxiv.org/abs/2101.07434v5","url_pdf":"https://arxiv.org/pdf/2101.07434v5.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":"caa-channelized-axial-attention-for-semantic","repo_url":"https://github.com/edwardyehuang/CAA","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":null,"task_name":"Relation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"axial-attention","method_name":"Axial Attention"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semantic-segmentation-on-coco-stuff-test","task":"Semantic Segmentation","dataset":"COCO-Stuff test","model":"CAA (Efficientnet-B7)","rank_in_archive_order":8,"of":21,"metrics":{"mIoU":"45.4%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-coco-stuff-test","task":"Semantic Segmentation","dataset":"COCO-Stuff test","model":"CAA (ResNet-101)","rank_in_archive_order":11,"of":21,"metrics":{"mIoU":"41.2%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-cityscapes","task":"Semantic Segmentation","dataset":"Cityscapes test","model":"CAA (ResNet-101)","rank_in_archive_order":26,"of":105,"metrics":{"Mean IoU (class)":"82.6%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-pascal-context","task":"Semantic Segmentation","dataset":"PASCAL Context","model":"CAA + Simple decoder (Efficientnet-B7)","rank_in_archive_order":12,"of":66,"metrics":{"mIoU":"60.5"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-pascal-context","task":"Semantic Segmentation","dataset":"PASCAL Context","model":"CAA (Efficientnet-B7)","rank_in_archive_order":14,"of":66,"metrics":{"mIoU":"60.1"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-pascal-context","task":"Semantic Segmentation","dataset":"PASCAL Context","model":"CAA (ResNet-101)","rank_in_archive_order":32,"of":66,"metrics":{"mIoU":"55.0"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2101.07434","atlas_url":"https://app.syntology.ai/?focus=2101.07434","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}