{"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/label-guided-knowledge-distillation-for","title":"Label-Guided Knowledge Distillation for Continual Semantic Segmentation on 2D Images and 3D Point Clouds","arxiv_id":null,"date":"2023-01-01","proceeding":"ICCV 2023 1","authors":["Ze Yang","Ruibo Li","Evan Ling","Chi Zhang","Yiming Wang","Dezhao Huang","Keng Teck Ma","Minhoe Hur","Guosheng Lin"],"abstract":"    Continual semantic segmentation (CSS) aims to extend an existing model to tackle unseen tasks while retaining its old knowledge. Naively fine-tuning the old model on new data leads to catastrophic forgetting. A common solution is knowledge distillation (KD), where the output distribution of the new model is regularized to be similar to that of the old model. However, in CSS, this is challenging because of the background shift issue. Existing KD-based CSS methods continue to suffer from confusion between the background and novel classes since they fail to establish a reliable class correspondence for distillation. To address this issue, we propose a new label-guided knowledge distillation (LGKD) loss, where the old model output is expanded and transplanted (with the guidance of the ground truth label) to form a semantically appropriate class correspondence with the new model output. Consequently, the useful knowledge from the old model can be effectively distilled into the new model without causing confusion. We conduct extensive experiments on two prevailing CSS benchmarks, Pascal-VOC and ADE20K, where our LGKD significantly boosts the performance of three competing methods, especially on novel mIoU by up to +76%, setting new state-of-the-art. Finally, to further demonstrate its generalization ability, we introduce the first CSS benchmark for 3D point cloud based on ScanNet, along with several re-implemented baselines for comparison. Experiments show that LGKD is versatile in both 2D and 3D modalities without requiring ad hoc design. Codes are available at https://github.com/Ze-Yang/LGKD.    ","url_abs":"http://openaccess.thecvf.com//content/ICCV2023/html/Yang_Label-Guided_Knowledge_Distillation_for_Continual_Semantic_Segmentation_on_2D_Images_ICCV_2023_paper.html","url_pdf":"http://openaccess.thecvf.com//content/ICCV2023/papers/Yang_Label-Guided_Knowledge_Distillation_for_Continual_Semantic_Segmentation_on_2D_Images_ICCV_2023_paper.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":"label-guided-knowledge-distillation-for","repo_url":"https://github.com/ze-yang/lgkd","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"continual-semantic-segmentation","task_name":"Continual Semantic Segmentation"},{"task_slug":"knowledge-distillation","task_name":"Knowledge Distillation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"hoc","method_name":"HOC"},{"method_slug":"knowledge-distillation","method_name":"Knowledge Distillation"},{"method_slug":"fail","method_name":"fail"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/continual-semantic-segmentation-on-ade20k","task":"Continual Semantic Segmentation","dataset":"ADE20K","model":"LGKD","rank_in_archive_order":1,"of":2,"metrics":{"mIoU":"37.5"},"uses_additional_data":false},{"leaderboard":"/sota/continual-semantic-segmentation-on-pascal-voc","task":"Continual Semantic Segmentation","dataset":"PASCAL VOC 2012","model":"LGKD","rank_in_archive_order":1,"of":1,"metrics":{"mIoU":"75.7"},"uses_additional_data":false},{"leaderboard":"/sota/continual-semantic-segmentation-on-scannet","task":"Continual Semantic Segmentation","dataset":"ScanNet","model":"LGKD","rank_in_archive_order":1,"of":1,"metrics":{"mIoU":"55.3"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}