Research map / 3D vision / Depth and stereo
ZeroStereo: Zero-shot Stereo Matching from Single Images
Xianqi Wang, Hao Yang, Gangwei Xu, Junda Cheng, Min Lin, Yong Deng, Jinliang Zang, Yurui Chen, Xin Yang
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 Adaptive Disparity Selection to the baseline stereo generation pipeline (ablation) | IGEV-Stereo · KITTI-15 All | EPE: 1.52 -> 1.24 | Improved |
| add Training-free Confidence Generation to the baseline (ablation) | IGEV-Stereo · KITTI-15 All | EPE: 1.52 -> 1.09 | Improved |
| add diffusion inpainting alone to the baseline (ablation) | IGEV-Stereo · KITTI-15 All | EPE: 1.52 -> 1.44 | No clear effect |
| add ADS together with inpainting (ablation) | IGEV-Stereo · KITTI-15 All | EPE: 1.44 -> 1.06 | Improved |
| add TCG on top of ADS and inpainting (ablation) | IGEV-Stereo · Midd-T H Noc | >2px error: 6.68 -> 7.11 | Hurt |
| add the ZeroStereo loss on top of ADS, inpainting and TCG (ablation) | IGEV-Stereo · ETH3D Noc | >1px error: 2.01 -> 1.90 | Improved |
| fine-tune the SDv2I inpainting model instead of using pre-trained weights (ablation) | IGEV-Stereo · Midd-T H Noc | >2px error: 7.21 -> 6.68 | Improved |
| use fine-tuned SDv2I instead of StereoDiffusion for synthesis | stereo image synthesis · 512x512 synthesis | time (s): 31.2 -> 1.9 | Improved |
| train on MfS35K instead of larger synthetic datasets such as FoundationStereo (ablation) | RAFT-Stereo · Midd-T | >2px error: 5.11 -> 4.45 | Improved |
| add confidence weighting 1-C to the photometric loss (ablation) | RAFT-Stereo · Midd-T H | >2px error: 4.95 -> 4.81 | Improved |
| add non-occlusion and inpainting masks to the confidence-weighted photometric loss (ablation) | RAFT-Stereo · KITTI-15 | >3px error: 4.76 -> 4.53 | Improved |
| replace NeRF-Stereo data with ZeroStereo data | RAFT-Stereo · DrivingStereo Foggy (F) | >3px error: 18.18 -> 8.66 | Improved |
Closest papers by meaning
- FoundationStereo: Zero-Shot Stereo Matching (CVPR 2025 · similarity 0.92)
- Stereo Anywhere: Robust Zero-Shot Deep Stereo Matching Even Where Either Stereo or Mono Fail (CVPR 2025 · similarity 0.90)
- Any-Stereo: Arbitrary Scale Disparity Estimation for Iterative Stereo Matching (AAAI 2024 · similarity 0.90)
- Zero-Shot Event-Intensity Asymmetric Stereo via Visual Prompting from Image Domain (NeurIPS 2024 · similarity 0.87)
- Mono2Stereo: A Benchmark and Empirical Study for Stereo Conversion (CVPR 2025 · similarity 0.87)
- MonSter: Marry Monodepth to Stereo Unleashes Power (CVPR 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.
