{"url":"/method/involution","slug":"involution","name":"Involution","full_name":"Involution","full_name_withheld":false,"description_markdown":"**Involution** is an atomic operation for deep neural networks that inverts the design principles of convolution. Involution kernels are distinct in the spatial extent but shared across channels. If involution kernels are parameterized as fixed-sized matrices like convolution kernels and updated using the back-propagation algorithm, the learned involution kernels are impeded from transferring between input images with variable resolutions. \r\n\r\nThe authors argue for two benefits of involution over convolution: (i) involution can summarize the context in a wider spatial arrangement, thus overcome the difficulty of modeling long-range interactions well; (ii) involution can adaptively allocate the weights over different positions, so as to prioritize the most informative visual elements in the spatial domain.","description_state":"present","introduced_year":null,"introduced_by":{"title":"Involution: Inverting the Inherence of Convolution for Visual Recognition","paper":"/paper/involution-inverting-the-inherence-of","first_author":"Duo Li","n_authors":8,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/involution-inverting-the-inherence-of"},"source":{"url":"https://arxiv.org/abs/2103.06255v2","title":"Involution: Inverting the Inherence of Convolution for Visual Recognition","url_on_a_paper_host":true},"code_snippet_url":null,"code_snippet_url_on_a_code_host":false,"categories":[{"area":"Computer Vision","area_id":"computer-vision","collection":"Image Feature Extractors","url":"/methods/category/image-feature-extractors","pwc_aliases":[]}],"n_papers_tagged":5,"archive_num_papers":5,"papers_newest_first":[{"paper":null,"title":"Involution and BSConv Multi-Depth Distillation Network for Lightweight Image Super-Resolution","date":"2025-03-18","arxiv_id":"2503.14779","n_code_links":0,"syntology":null},{"paper":null,"title":"LAM-YOLO: Drones-based Small Object Detection on Lighting-Occlusion Attention Mechanism YOLO","date":"2024-11-01","arxiv_id":"2411.00485","n_code_links":0,"syntology":null},{"paper":null,"title":"GIU-GANs: Global Information Utilization for Generative Adversarial Networks","date":"2022-01-25","arxiv_id":"2201.10471","n_code_links":0,"syntology":null},{"paper":null,"title":"Automatic Modulation Classification Using Involution Enabled Residual Networks","date":"2021-08-23","arxiv_id":"2108.10001","n_code_links":0,"syntology":null},{"paper":"/paper/involution-inverting-the-inherence-of","title":"Involution: Inverting the Inherence of Convolution for Visual Recognition","date":"2021-03-10","arxiv_id":"2103.06255","n_code_links":13,"syntology":{"ran":7,"of":9,"unverified":2,"pointer_only":4}}],"papers_shown":5,"tasks":[{"task":"/task/classification-1","name":"Classification","papers":1},{"task":"/task/computational-efficiency","name":"Computational Efficiency","papers":1},{"task":"/task/image-classification","name":"Image Classification","papers":1},{"task":"/task/image-generation","name":"Image Generation","papers":1},{"task":"/task/image-super-resolution","name":"Image Super-Resolution","papers":1},{"task":"/task/object-detection","name":"Object Detection","papers":1},{"task":"/task/ssim","name":"SSIM","papers":1},{"task":"/task/small-object-detection","name":"Small Object Detection","papers":1},{"task":"/task/super-resolution","name":"Super-Resolution","papers":1},{"task":"/task/object-detection-1","name":"object-detection","papers":1}],"tasks_shown":10,"n_tasks":10,"usage_by_year":[{"year":"2021","papers":2},{"year":"2022","papers":1},{"year":"2024","papers":1},{"year":"2025","papers":1}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/involution"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}