WAFT-Stereo: Warping-Alone Field Transforms for Stereo Matching

TL;DR

WAFT-Stereo introduces a warping-based stereo matching method that eliminates the need for cost volumes, achieving top performance and efficiency on multiple benchmarks.

Contribution

The paper presents a novel warping-alone approach for stereo matching that outperforms cost volume methods in accuracy and speed.

Findings

Ranks first on ETH3D, Middlebury, and KITTI benchmarks.

Reduces zero-shot error by 81% on ETH3D.

Is 1.8-6.7x faster than competitive methods.

Abstract

We introduce WAFT-Stereo, a simple and effective warping-based method for stereo matching. WAFT-Stereo demonstrates that cost volumes, a common design used in many leading methods, are not necessary for strong performance and can be replaced by warping with improved efficiency. WAFT-Stereo ranks first on ETH3D (BP-0.5), Middlebury (RMSE), and KITTI (all metrics), reducing the zero-shot error by 81% on ETH3D, while being 1.8-6.7x faster than competitive methods. Code and model weights are available at https://github.com/princeton-vl/WAFT-Stereo.