Research map / 3D vision / Depth and stereo
NVDS+: Towards Efficient and Versatile Neural Stabilizer for Video Depth Estimation
Yiran Wang, Min Shi, Jiaqi Li, Chaoyi Hong, Zihao Huang, Juewen Peng, Zhiguo Cao, Jianming Zhang, Ke Xian, Guosheng Lin
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 bidirectional averaging over forward-only inference (ablation) | NVDS+ Large (DPT-Large) · VDW test | OPW: 0.165 -> 0.147 | Improved |
| add flow-guided consistency fusion on top of averaging (ablation) | NVDS+ Large (DPT-Large) · VDW test | OPW: 0.147 -> 0.129 | Improved |
| add flow-guided consistency fusion on top of averaging (ablation) | NVDS+ Large (MiDaS-v2.1-Large) · VDW test | OPW: 0.180 -> 0.148 | Improved |
| add temporal loss during training (ablation) | NVDS+ Large · VDW subset | OPW: 0.303 -> 0.216 | Improved |
| add temporal loss during training (ablation) | NVDS+ Large · VDW subset | delta1: 0.627 -> 0.625 | No clear effect |
| use three reference frames instead of one (ablation) | NVDS+ Large · VDW subset | OPW: 0.272 -> 0.216 | Improved |
| use four reference frames instead of three (ablation) | NVDS+ Large · VDW subset | OPW: 0.216 -> 0.224 | No clear effect |
| increase inter-frame interval from 1 to 5 (ablation) | NVDS+ Large · VDW subset | OPW: 0.216 -> 0.246 | Hurt |
| add the stabilization network instead of a multi-frame baseline (ablation) | DPT-Large · VDW subset | OPW: 0.471 -> 0.216 | Improved |
| train on VDW instead of NYUDV2 (ablation) | NVDS+ · Sintel (zero-shot) | delta1: 0.527 -> 0.591 | Improved |
| train on VDW instead of IRS+TartanAir (ablation) | NVDS+ · Sintel (zero-shot) | OPW: 0.489 -> 0.424 | Improved |
| pretrain on VDW before finetuning on NYUDV2 (ablation) | NVDS+ Large (DPT-Large) · NYUDV2 | delta1: 0.931 -> 0.950 | Improved |
Closest papers by meaning
- Video Depth Anything: Consistent Depth Estimation for Super-Long Videos (CVPR 2025 · similarity 0.90)
- CH3Depth: Efficient and Flexible Depth Foundation Model with Flow Matching (CVPR 2025 · similarity 0.89)
- Learning Temporally Consistent Video Depth from Video Diffusion Priors (CVPR 2025 · similarity 0.88)
- Depth Any Video with Scalable Synthetic Data (ICLR 2025 · similarity 0.88)
- DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos (CVPR 2025 · similarity 0.87)
- Align3R: Aligned Monocular Depth Estimation for Dynamic Videos (CVPR 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.
