WAFT-Stereo: Warping-Alone Field Transforms for Stereo Matching
This addresses the efficiency and accuracy problem in stereo vision for applications like robotics and autonomous driving, representing a novel paradigm shift.
The paper tackled stereo matching by introducing WAFT-Stereo, a warping-based method that eliminates cost volumes, achieving top rankings on ETH3D, KITTI, and Middlebury benchmarks with an 81% reduction in zero-shot error on ETH3D and 1.8-6.7x speed improvements.
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, KITTI and Middlebury public benchmarks, reducing the zero-shot error by 81% on ETH3D benchmark, while being 1.8-6.7x faster than competitive methods. Code and model weights are available at https://github.com/princeton-vl/WAFT-Stereo.