FoundationStereo: Zero-Shot Stereo Matching

TL;DR

FoundationStereo introduces a large synthetic dataset and novel architecture components to achieve strong zero-shot generalization in stereo depth estimation, surpassing previous models in robustness and accuracy across diverse domains.

Contribution

The paper presents a new foundation model for stereo matching that leverages a large synthetic dataset and innovative network design for improved zero-shot generalization.

Findings

Achieves state-of-the-art zero-shot stereo matching performance.

Demonstrates robustness across multiple real-world domains.

Introduces a scalable architecture with monocular priors and context reasoning.

Abstract

Tremendous progress has been made in deep stereo matching to excel on benchmark datasets through per-domain fine-tuning. However, achieving strong zero-shot generalization - a hallmark of foundation models in other computer vision tasks - remains challenging for stereo matching. We introduce FoundationStereo, a foundation model for stereo depth estimation designed to achieve strong zero-shot generalization. To this end, we first construct a large-scale (1M stereo pairs) synthetic training dataset featuring large diversity and high photorealism, followed by an automatic self-curation pipeline to remove ambiguous samples. We then design a number of network architecture components to enhance scalability, including a side-tuning feature backbone that adapts rich monocular priors from vision foundation models to mitigate the sim-to-real gap, and long-range context reasoning for effective cost volume filtering. Together, these components lead to strong robustness and accuracy across domains, establishing a new standard in zero-shot stereo depth estimation. Project page: https://nvlabs.github.io/FoundationStereo/