Uncertainty Guided Adaptive Warping for Robust and Efficient Stereo Matching

TL;DR

This paper introduces a novel Uncertainty Guided Adaptive Correlation (UGAC) module for stereo matching, enabling a fixed model to adapt dynamically across datasets, improving robustness and efficiency without retraining.

Contribution

The paper presents a new UGAC module that adaptively adjusts correlation sampling based on uncertainty, enhancing stereo matching robustness across diverse datasets.

Findings

Achieves state-of-the-art results on ETH3D, KITTI, and Middlebury datasets.

Outperforms existing methods with a lightweight model of only 0.6 million parameters.

Demonstrates robustness without retraining across different datasets.

Abstract

Correlation based stereo matching has achieved outstanding performance, which pursues cost volume between two feature maps. Unfortunately, current methods with a fixed model do not work uniformly well across various datasets, greatly limiting their real-world applicability. To tackle this issue, this paper proposes a new perspective to dynamically calculate correlation for robust stereo matching. A novel Uncertainty Guided Adaptive Correlation (UGAC) module is introduced to robustly adapt the same model for different scenarios. Specifically, a variance-based uncertainty estimation is employed to adaptively adjust the sampling area during warping operation. Additionally, we improve the traditional non-parametric warping with learnable parameters, such that the position-specific weights can be learned. We show that by empowering the recurrent network with the UGAC module, stereo matching can be exploited more robustly and effectively. Extensive experiments demonstrate that our method achieves state-of-the-art performance over the ETH3D, KITTI, and Middlebury datasets when employing the same fixed model over these datasets without any retraining procedure. To target real-time applications, we further design a lightweight model based on UGAC, which also outperforms other methods over KITTI benchmarks with only 0.6 M parameters.