UWStereo: A Large Synthetic Dataset for Underwater Stereo Matching

TL;DR

This paper introduces UWStereo, a large synthetic underwater stereo dataset with detailed annotations, designed to improve deep learning models for underwater stereo matching, and proposes new strategies to enhance model generalization across domains.

Contribution

The paper presents UWStereo, a comprehensive synthetic underwater stereo dataset with diverse scenes and annotations, and introduces novel training strategies to improve model generalization in underwater environments.

Findings

Existing models struggle to generalize to new underwater domains.

UWStereo dataset outperforms existing datasets in scale and realism.

Proposed methods improve cross-domain stereo matching performance.

Abstract

Despite recent advances in stereo matching, the extension to intricate underwater settings remains unexplored, primarily owing to: 1) the reduced visibility, low contrast, and other adverse effects of underwater images; 2) the difficulty in obtaining ground truth data for training deep learning models, i.e. simultaneously capturing an image and estimating its corresponding pixel-wise depth information in underwater environments. To enable further advance in underwater stereo matching, we introduce a large synthetic dataset called UWStereo. Our dataset includes 29,568 synthetic stereo image pairs with dense and accurate disparity annotations for left view. We design four distinct underwater scenes filled with diverse objects such as corals, ships and robots. We also induce additional variations in camera model, lighting, and environmental effects. In comparison with existing underwater datasets, UWStereo is superior in terms of scale, variation, annotation, and photo-realistic image quality. To substantiate the efficacy of the UWStereo dataset, we undertake a comprehensive evaluation compared with nine state-of-the-art algorithms as benchmarks. The results indicate that current models still struggle to generalize to new domains. Hence, we design a new strategy that learns to reconstruct cross domain masked images before stereo matching training and integrate a cross view attention enhancement module that aggregates long-range content information to enhance the generalization ability.