Research map / 3D vision / Depth and stereo
Disentangling Object Motion and Occlusion for Unsupervised Multi-Frame Monocular Depth
Ziyue Feng, Liang Yang, Longlong Jing, Haiyan Wang, YingLi Tian, Bing Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Dynamic Object Motion Disentanglement (ablation) | ManyDepth-style multi-frame model · Cityscapes test | Abs Rel: 0.114 -> 0.110 | Improved |
| add occlusion-aware loss switching alone (ablation) | DOMD model · Cityscapes test | Abs Rel: 0.110 -> 0.110 | No clear effect |
| add occlusion masking alone (ablation) | DOMD model · Cityscapes test | Abs Rel: 0.110 -> 0.110 | No clear effect |
| add both switching and masking in re-projection loss (ablation) | DOMD model · Cityscapes test | Abs Rel: 0.110 -> 0.108 | Improved |
| add occlusion-aware cost volume (ablation) | DOMD model · Cityscapes test | Abs Rel: 0.110 -> 0.108 | Improved |
| add occlusion-aware cost volume and loss together (ablation) | DOMD model · Cityscapes test | Abs Rel: 0.110 -> 0.107 | Improved |
| add dynamic object cycle consistency (unfreeze depth prior) (ablation) | DynamicDepth · Cityscapes test | Abs Rel: 0.107 -> 0.103 | Improved |
| add dynamic object cycle consistency (unfreeze depth prior) (ablation) | DynamicDepth · Cityscapes test | Sq Rel: 1.121 -> 1.000 | Improved |
| replace ManyDepth with full DynamicDepth | ManyDepth · Cityscapes test | Abs Rel: 0.114 -> 0.103 | Improved |
| replace ManyDepth with DynamicDepth | ManyDepth · KITTI Eigen (original GT) | Abs Rel: 0.098 -> 0.096 | Improved |
| replace ManyDepth with DynamicDepth on dynamic regions | ManyDepth · KITTI (dynamic object areas) | Abs Rel: 0.175 -> 0.150 | Improved |
| multi-frame DynamicDepth vs its single-frame depth prior | Our Depth Prior · Cityscapes (dynamic object areas) | Abs Rel: 0.137 -> 0.129 | Improved |
Closest papers by meaning
- AdaDepth: Exploiting Inherent Scene Information for Self-Supervised Depth Estimation in Dynamic Scenes (AAAI 2026 · similarity 0.91)
- Mining Supervision for Dynamic Regions in Self-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.91)
- ProDepth: Boosting Self-Supervised Multi-Frame Monocular Depth with Probabilistic Fusion (ECCV 2024 · similarity 0.90)
- Learning Monocular Depth in Dynamic Scenes via Instance-Aware Projection Consistency (AAAI 2021 · similarity 0.90)
- The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth (CVPR 2021 · similarity 0.90)
- From-Ground-To-Objects: Coarse-to-Fine Self-supervised Monocular Depth Estimation of Dynamic Objects with Ground Contact Prior (CVPR 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
