{"url":"/method/k-net","slug":"k-net","name":"K-Net","full_name":"K-Net","full_name_withheld":false,"description_markdown":"**K-Net** is a framework for unified semantic and instance segmentation that segments both instances and semantic categories consistently by a group of learnable kernels, where each kernel is responsible for generating a mask for either a potential instance or a stuff class. It begins with a set of kernels that are randomly initialized, and learns the kernels in accordance to the segmentation targets at hand, namely, semantic kernels for semantic categories and instance kernels for instance identities. A simple combination of semantic kernels and instance kernels allows panoptic segmentation naturally. In the forward pass, the kernels perform [convolution](https://paperswithcode.com/method/convolution) on the image features to obtain the corresponding segmentation predictions.\r\n\r\nK-Net is formulated so that it dynamically updates the kernels to make them conditional to their activations on the image. Such a content-aware mechanism is crucial to ensure that each kernel, especially an instance kernel, responds accurately to varying objects in an image. Through applying this adaptive kernel update strategy iteratively, K-Net significantly improves the discriminative ability of the kernels and boosts the final segmentation performance. It is noteworthy that this strategy universally applies to kernels for all the segmentation tasks.\r\n\r\nIt also utilises a bipartite matching strategy to assign learning targets for each kernel. This training approach is advantageous to conventional training strategies as it builds a one-to-one mapping between kernels and instances in an image. It thus resolves the problem of dealing with a varying number of instances in an image. In addition, it is purely mask-driven without involving boxes. Hence, K-Net is naturally [NMS](https://paperswithcode.com/method/non-maximum-suppression)-free and box-free, which is appealing to real-time applications.","description_state":"present","introduced_year":null,"introduced_by":{"title":"K-Net: Towards Unified Image Segmentation","paper":"/paper/k-net-towards-unified-image-segmentation","first_author":"Wenwei Zhang","n_authors":4,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/k-net-towards-unified-image-segmentation"},"source":{"url":"https://arxiv.org/abs/2106.14855v2","title":"K-Net: Towards Unified Image Segmentation","url_on_a_paper_host":true},"code_snippet_url":"https://github.com/ZwwWayne/K-Net","code_snippet_url_on_a_code_host":true,"categories":[{"area":"Computer Vision","area_id":"computer-vision","collection":"Instance Segmentation Models","url":"/methods/category/instance-segmentation-models","pwc_aliases":[]},{"area":"Computer Vision","area_id":"computer-vision","collection":"Semantic Segmentation Models","url":"/methods/category/semantic-segmentation-models","pwc_aliases":["segmentation-models"]}],"n_papers_tagged":6,"archive_num_papers":6,"papers_newest_first":[{"paper":null,"title":"Segmenting objects with Bayesian fusion of active contour models and convnet priors","date":"2024-10-09","arxiv_id":"2410.07421","n_code_links":0,"syntology":null},{"paper":"/paper/rt-k-net-revisiting-k-net-for-real-time","title":"RT-K-Net: Revisiting K-Net for Real-Time Panoptic Segmentation","date":"2023-05-02","arxiv_id":"2305.01255","n_code_links":1,"syntology":null},{"paper":"/paper/tube-link-a-flexible-cross-tube-baseline-for","title":"Tube-Link: A Flexible Cross Tube Framework for Universal Video Segmentation","date":"2023-03-22","arxiv_id":"2303.12782","n_code_links":1,"syntology":null},{"paper":"/paper/video-k-net-a-simple-strong-and-unified","title":"Video K-Net: A Simple, Strong, and Unified Baseline for Video Segmentation","date":"2022-04-10","arxiv_id":"2204.04656","n_code_links":1,"syntology":null},{"paper":null,"title":"Dual-Domain Reconstruction Networks with V-Net and K-Net for fast MRI","date":"2022-03-11","arxiv_id":"2203.05725","n_code_links":0,"syntology":null},{"paper":"/paper/k-net-towards-unified-image-segmentation","title":"K-Net: Towards Unified Image Segmentation","date":"2021-06-28","arxiv_id":"2106.14855","n_code_links":1,"syntology":{"ran":0,"of":1,"unverified":1,"pointer_only":0}}],"papers_shown":6,"tasks":[{"task":"/task/instance-segmentation","name":"Instance Segmentation","papers":4},{"task":"/task/segmentation","name":"Segmentation","papers":4},{"task":"/task/semantic-segmentation","name":"Semantic Segmentation","papers":4},{"task":"/task/panoptic-segmentation","name":"Panoptic Segmentation","papers":3},{"task":"/task/image-segmentation","name":"Image Segmentation","papers":2},{"task":"/task/video-instance-segmentation","name":"Video Instance Segmentation","papers":2},{"task":"/task/video-panoptic-segmentation","name":"Video Panoptic Segmentation","papers":2},{"task":"/task/video-segmentation","name":"Video Segmentation","papers":2},{"task":"/task/video-semantic-segmentation","name":"Video Semantic Segmentation","papers":2},{"task":"/task/contrastive-learning","name":"Contrastive Learning","papers":1},{"task":"/task/decoder","name":"Decoder","papers":1},{"task":null,"name":"GPU","papers":1},{"task":"/task/image-reconstruction","name":"Image Reconstruction","papers":1},{"task":"/task/scene-parsing","name":"Scene Parsing","papers":1}],"tasks_shown":14,"n_tasks":14,"usage_by_year":[{"year":"2021","papers":1},{"year":"2022","papers":2},{"year":"2023","papers":2},{"year":"2024","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/k-net"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}