Research map / 3D vision / Depth and stereo
Instance-Level Video Depth in Groups Beyond Occlusions
Yuan Liang, Yang Zhou, Ziming Sun, Tianyi Xiang, Guiqing Li, Shengfeng He
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 Holistic Depth Initialization (ablation) | DepthAnything V2 + DPT baseline · GID test set | RMS: 0.4357 -> 0.4188 | Improved |
| add Holistic Depth Initialization (ablation) | DepthAnything V2 + DPT baseline · GID test set | REL: 0.0524 -> 0.0506 | Improved |
| add Instance-Aware Depth Rectification (ablation) | Baseline+H · GID test set | RMS: 0.4188 -> 0.397 | Improved |
| add Instance-Aware Depth Rectification (ablation) | Baseline+H · GID test set | REL: 0.0506 -> 0.045 | Improved |
| add Instance-Aware Depth Rectification (ablation) | Baseline+H · GID test set | sigma1 accuracy: 0.9784 -> 0.983 | Improved |
| use 2 meter depth range partition instead of 1 meter (ablation) | Holistic initialization · GID test set | RMS: 0.426 -> 0.419 | Improved |
| use 3 meter depth range partition instead of 2 meter (ablation) | Holistic initialization · GID test set | RMS: 0.419 -> 0.431 | Hurt |
| add Ldist to Lobj (ablation) | Instance-aware rectification · GID test set | RMS: 0.407 -> 0.397 | Improved |
| add Lobj to Ldist (ablation) | Instance-aware rectification · GID test set | RMS: 0.412 -> 0.397 | Improved |
| add Ldist to Lobj (ablation) | Instance-aware rectification · GID test set | sigma1 accuracy: 0.974 -> 0.983 | Improved |
| use InstanceDepth instead of DepthAnything V2 on static indoor data | InstanceDepth · NYU Depth V2 test | RMS: 0.206 -> 0.202 | No clear effect |
| use InstanceDepth instead of DepthAnything V2 | InstanceDepth · GID test set | RMS: 0.435 -> 0.397 | Improved |
Closest papers by meaning
- GeoDepth: From Point-to-Depth to Plane-to-Depth Modeling for Self-Supervised Monocular Depth Estimation (CVPR 2025 · similarity 0.89)
- Video Depth Anything: Consistent Depth Estimation for Super-Long Videos (CVPR 2025 · similarity 0.89)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.88)
- Seurat: From Moving Points to Depth (CVPR 2025 · similarity 0.87)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.87)
- Depth Any Video with Scalable Synthetic Data (ICLR 2025 · similarity 0.87)
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.
