Research map / 3D vision / Depth and stereo
RM-Depth: Unsupervised Learning of Recurrent Monocular Depth in Dynamic Scenes
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use RM-Depth instead of PackNet | Monocular depth estimation · KITTI Eigen (K training) | AbsRel: 0.111 -> 0.107 | Helped |
| Use RM-Depth instead of Lee et al. | Monocular depth estimation · Cityscapes test (CS training) | AbsRel: 0.111 -> 0.090 | Helped |
| Add residual upsampling (ablation) | RM-Depth · KITTI ablation test split | AbsRel: 0.1097 -> 0.1081 | Helped |
| Add recurrent modulation units (ablation) | RM-Depth · KITTI ablation test split | AbsRel: 0.1167 -> 0.1081 | Helped |
| Add adaptive modulation within RMU (ablation) | RM-Depth · KITTI ablation test split | AbsRel: 0.1165 -> 0.1081 | Helped |
| Add warping for object-motion estimation (ablation) | RM-Depth · Cityscapes ablation test split | AbsRel: 0.0933 -> 0.0903 | Helped |
| Add outlier-aware motion regularization (ablation) | RM-Depth · Cityscapes ablation test split | AbsRel: 0.0995 -> 0.0903 | Helped |
| Use outlier-aware regularization instead of sparsity loss (ablation) | RM-Depth · Cityscapes ablation test split | AbsRel: 0.1066 -> 0.0903 | Helped |
| Add object motion estimation (ablation) | RM-Depth · Cityscapes ablation test split | AbsRel: 0.1174 -> 0.0903 | Helped |
| Use 13 rather than 3 RMUs (ablation) | RM-Depth · KITTI ablation test split | AbsRel: 0.1161 -> 0.1081 | Helped |
| Use 13 rather than 3 RMUs (ablation) | RM-Depth · KITTI ablation test split | Runtime: 14.99 -> 24.78 | Hurt |
| Enable warping (ablation) | RM-Depth optical-flow estimation · KITTI 2015 optical flow | AEE All: 10.45 -> 8.16 | Helped |
Closest papers by meaning
- Dynamo-Depth: Fixing Unsupervised Depth Estimation for Dynamical Scenes (NeurIPS 2023 · similarity 0.93)
- Depth Prediction without the Sensors: Leveraging Structure for Unsupervised Learning from Monocular Videos (AAAI 2019 · similarity 0.91)
- Learning Monocular Depth in Dynamic Scenes via Instance-Aware Projection Consistency (AAAI 2021 · similarity 0.90)
- Mining Supervision for Dynamic Regions in Self-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.90)
- Seurat: From Moving Points to Depth (CVPR 2025 · similarity 0.88)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.88)
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.