{"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/three-ways-to-improve-semantic-segmentation","title":"Three Ways to Improve Semantic Segmentation with Self-Supervised Depth Estimation","arxiv_id":"2012.10782","date":"2020-12-19","proceeding":"CVPR 2021 1","authors":["Lukas Hoyer","Dengxin Dai","Yuhua Chen","Adrian Köring","Suman Saha","Luc van Gool"],"abstract":"Training deep networks for semantic segmentation requires large amounts of labeled training data, which presents a major challenge in practice, as labeling segmentation masks is a highly labor-intensive process. To address this issue, we present a framework for semi-supervised semantic segmentation, which is enhanced by self-supervised monocular depth estimation from unlabeled image sequences. In particular, we propose three key contributions: (1) We transfer knowledge from features learned during self-supervised depth estimation to semantic segmentation, (2) we implement a strong data augmentation by blending images and labels using the geometry of the scene, and (3) we utilize the depth feature diversity as well as the level of difficulty of learning depth in a student-teacher framework to select the most useful samples to be annotated for semantic segmentation. We validate the proposed model on the Cityscapes dataset, where all three modules demonstrate significant performance gains, and we achieve state-of-the-art results for semi-supervised semantic segmentation. The implementation is available at https://github.com/lhoyer/improving_segmentation_with_selfsupervised_depth.","url_abs":"https://arxiv.org/abs/2012.10782v2","url_pdf":"https://arxiv.org/pdf/2012.10782v2.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":"three-ways-to-improve-semantic-segmentation","repo_url":"https://github.com/lhoyer/improving_segmentation_with_selfsupervised_depth","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"depth-estimation","task_name":"Depth Estimation"},{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"monocular-depth-estimation","task_name":"Monocular Depth Estimation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"semi-supervised-semantic-segmentation","task_name":"Semi-Supervised Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-3","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 100 samples labeled","model":"SegSDE (MTL decoder with ResNet101, ImageNet pretrained, unlabeled image sequences)","rank_in_archive_order":5,"of":13,"metrics":{"Validation mIoU":"62.09%"},"uses_additional_data":true},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-2","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 12.5% labeled","model":"SegSDE (MTL decoder with ResNet101, ImageNet pretrained, unlabeled image sequences)","rank_in_archive_order":23,"of":33,"metrics":{"Validation mIoU":"68.01%"},"uses_additional_data":true},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-1","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 25% labeled","model":"SegSDE (MTL decoder with ResNet101, ImageNet pretrained, unlabeled image sequences)","rank_in_archive_order":21,"of":30,"metrics":{"Validation mIoU":"69.38%"},"uses_additional_data":true}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2012.10782","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}