Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
[CVPR 2025] DEFOM-Stereo: Depth foundation model based stereo matching
| Date | Stars |
|---|---|
| 2026-07-31 | 286 |
| 2026-08-05 | 288 |
| 2026-08-06 | 288 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# DEFOM-Stereo [CVPR 2025]
The Official Pytorch Implementation for
> [**DEFOM-Stereo: Depth Foundation Model Based Stereo Matching**](https://arxiv.org/abs/2501.09466)
>
> Authors: Hualie Jiang, Zhiqiang Lou, Laiyan Ding, Rui Xu, Minglang Tan, Wenjie Jiang and Rui Huang
# Abstract
Stereo matching is a key technique for metric depth estimation in computer vision and robotics.
Real-world challenges like occlusion and non-texture hinder accurate disparity estimation from binocular matching cues. Recently, monocular relative depth estimation has shown remarkable generalization using vision foundation models. Thus, to facilitate robust stereo matching with monocular depth cues, we incorporate a robust monocular relative depth model into the recurrent stereo-matching framework, building a new framework for depth foundation model-based stereo-matching, DEFOM-Stereo.
In the feature extraction stage, we construct the combined context and matching feature encoder by integrating features from conventional CNNs and DEFOM. In the update stage, we use the depth predicted by DEFOM to initialize the recurrent disparity and introduce a scale update module to refine the disparity at the correct scale. DEFOM-Stereo is verified to have much stronger zero-shot generalization compared with SOTA methods. Moreover, DEFOM-Stereo achieves top performance on the KITTI 2012, KITTI 2015, Middlebury, and ETH3D benchmarks, ranking $1^{st}$ on many metrics. In the joint evaluation under the robust vision challenge, our model simultaneously outperforms previous models on the individual benchmarks, further demonstrating its outstanding capabilities.
# Pipeline
<p align="center">
<img src='assets/framework.svg' width=980>
</p>
- We propose **a novel recurrent stereo-matching framework incorporating monocular depth cues** from a depth foundation model to improve robustness.
- We develop a simple technique that utilizes **pre-trained DEFOM features** to construct stronger **combined feature and context encoders**.
- We invent a **recurrent scale update** module empowered with the **scale lookup**, serving to recover accurate pixel-wise scales for the coarse DEFOM depth.
# Zero-Shot Perfomance
<p align="center">
<img src='assets/zeroshot.jpg' width=1000>
</p>
# Benchmark Performance
<p align="center">
<img src='assets/benchmark.jpg' width=800>
</p>
# Robust Vision Challange
<p align="center">
<img src='assets/rvc.svg' width=1200>
</p>
# Preparation
### Installation
Create the environment
```bash
conda env create -f environment.yaml
conda activate defomstereo
pip install -r requirements.txt
```
### Datasets
The project requires the follow datasets:
<table style="border-collapse: collapse; width: 80%;">
<tr>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://www.cvlibs.net/datasets/kitti/eval_stereo_flow.php?benchmark=stereo" target="_blank">KITTI-2012</a></td>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://www.cvlibs.net/datasets/kitti/eval_scene_flow.php?benchmark=stereo" target="_blank">KITTI-2015</a></td>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://vision.middlebury.edu/stereo/submit3/" target="_blank">Middlebury</a></td>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://www.eth3d.net/datasets" target="_blank">ETH3D</a></td>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://github.com/YuhuaXu/StereoDataset" target="_blank">InStereo2K</a></td>
</tr>
<tr>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://europe.naverlabs.com/proxy-virtual-worlds-vkitti-2/" target="_blank">Virtual KITTI 2</a></td>
<td style="border: 1px solid #000; padding: 8px;"><a class="custom-link" href="https://lmb.informatik.uni-freiburg.de/resources/datasets/SceneFlowDatasets.en.html" target="_blank">SceneFlExcerpt of 8,383 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:1bbddcbab2932f48, llm:topics: cvpr2025, depth-estimation, monocular-depth, stereo-matching, stereo-vision; description: 'DEFOM-Stereo: Depth foundation model based stereo matching'
matched fp:1bbddcbab2932f48, llm:topics: cvpr2025, depth-estimation, monocular-depth, stereo-matching, stereo-vision; description: 'DEFOM-Stereo: Depth foundation model based stereo matching'
matched fp:1bbddcbab2932f48, llm:topics: cvpr2025, depth-estimation, monocular-depth, stereo-matching, stereo-vision; description: 'DEFOM-Stereo: Depth foundation model based stereo matching'