Research map / 3D vision / Depth and stereo
The Midas Touch for Metric Depth
Yu Ma, Zizhan Guo, Zuyi Xiong, Haoran Zhang, Yi Feng, Hongbo Zhao, Hanli Wang, Rui Fan
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 |
|---|---|---|---|
| use inverse-depth proxy instead of least squares in depth domain (ablation) | per-segment calibration · KITTI | RMSE: 7.013 -> 6.782 | Improved |
| use least squares instead of median alignment (ablation) | per-segment calibration · KITTI | RMSE: 10.891 -> 7.013 | Improved |
| add sparse graph optimization instead of global scale recovery (ablation) | segment-wise recovery · KITTI | RMSE: 2.521 -> 2.232 | Improved |
| add sparse graph optimization instead of global scale recovery (ablation) | segment-wise recovery · VOID | RMSE: 0.554 -> 0.459 | Improved |
| add discontinuity-aware geodesic cost (ablation) | dynamic programming refinement · KITTI | RMSE: 2.618 -> 2.049 | Improved |
| use polynomial basis instead of B-spline (ablation) | dynamic programming refinement · KITTI | RMSE: 2.112 -> 2.049 | Improved |
| increase iterations k from 3 to 5 (ablation) | dynamic programming refinement · KITTI | RMSE: 2.028 -> 1.913 | Improved |
| increase iterations k from 3 to 5 (ablation) | dynamic programming refinement · VOID | RMSE: 0.366 -> 0.391 | Hurt |
| add MTD to DepthAnythingV2 relative depth | DepthAnythingV2 · KITTI | AbsRel: 0.080 -> 0.022 | Improved |
| add MTD to MiDaS relative depth | MiDaS · DIODE | AbsRel: 0.266 -> 0.113 | Improved |
| add MTD to Droid SLAM instead of LiDAR only | Droid · KITTI Odometry Seq 06 | ATE-RMSE: 104.552 -> 12.194 | Improved |
| use MTD instead of least-squares alignment on VGGT depth | VGGT · Multi-view stereo | AbsRel: 0.108 -> 0.054 | Improved |
Closest papers by meaning
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
- Metric Depth Estimation from Arbitrarily Degraded Low-Resolution Depth Prompts (NeurIPS 2026 · similarity 0.88)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.87)
- SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation (CVPR 2025 · similarity 0.87)
- GeoDepth: From Point-to-Depth to Plane-to-Depth Modeling for Self-Supervised Monocular Depth Estimation (CVPR 2025 · similarity 0.87)
- Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Prior (AAAI 2025 · similarity 0.86)
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.
