{"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/2d-semantic-guided-semantic-scene-completion","title":"2D Semantic-Guided Semantic Scene Completion","arxiv_id":null,"date":"2024-09-12","proceeding":"International Journal of Computer Vision (IJCV) 2024 9","authors":["Xianzhu Liu","Haozhe Xie","Shengping Zhang","Hongxun Yao","Rongrong Ji","Liqiang Nie","DaCheng Tao"],"abstract":"Semantic scene completion (SSC) aims to simultaneously perform scene completion (SC) and predict semantic categories\r\nof a 3D scene from a single depth and/or RGB image. Most existing SSC methods struggle to handle complex regions with\r\nmultiple objects close to each other, especially for objects with reflective or dark surfaces. This primarily stems from two\r\nchallenges: (1) the loss of geometric information due to the unreliability of depth values from sensors, and (2) the potential for\r\nsemantic confusion when simultaneously predicting 3D shapes and semantic labels. To address these problems, we propose a\r\nSemantic-guided Semantic Scene Completion framework, dubbed SG-SSC, which involves Semantic-guided Fusion (SGF)\r\nand Volume-guided Semantic Predictor (VGSP). Guided by 2D semantic segmentation maps, SGF adaptively fuses RGB\r\nand depth features to compensate for the missing geometric information caused by the missing values in depth images, thus\r\nperforming more robustly to unreliable depth information. VGSP exploits the mutual benefit between SC and SSC tasks,\r\nmaking SSC more focused on predicting the categories of voxels with high occupancy probabilities and also allowing SC\r\nto utilize semantic priors to better predict voxel occupancy. Experimental results show that SG-SSC outperforms existing\r\nstate-of-the-art methods on the NYU, NYUCAD, and SemanticKITTI datasets. Models and code are available at https://\r\ngithub.com/aipixel/SG-SSC.","url_abs":"https://doi.org/10.1007/s11263-024-02244-y","url_pdf":"https://trebuchet.public.springernature.app/get_content/fc0c39ab-cba9-456c-afe7-aaaafb78c09e?utm_source=rct_congratemailt&utm_medium=email&utm_campaign=nonoa_20241003&utm_content=10.1007/s11263-024-02244-y","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":"2d-semantic-guided-semantic-scene-completion","repo_url":"https://github.com/aipixel/SG-SSC","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"2d-semantic-segmentation","task_name":"2D Semantic Segmentation"},{"task_slug":"3d-semantic-scene-completion","task_name":"3D Semantic Scene Completion"},{"task_slug":"missing-values","task_name":"Missing Values"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-semantic-scene-completion-on-nyuv2","task":"3D Semantic Scene Completion","dataset":"NYUv2","model":"SG-SSC","rank_in_archive_order":1,"of":28,"metrics":{"mIoU":"55.4"},"uses_additional_data":false},{"leaderboard":"/sota/3d-semantic-scene-completion-on-semantickitti","task":"3D Semantic Scene Completion","dataset":"SemanticKITTI","model":"SG-SSC","rank_in_archive_order":17,"of":20,"metrics":{"mIoU":"12.9"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}