MIDAS: Modeling Ground-Truth Distributions with Dark Knowledge for Domain Generalized Stereo Matching
This work addresses domain-specific biases in stereo matching for practical applications in diverse real-world scenarios, representing an incremental improvement over existing methods.
The paper tackles the problem of domain generalization in stereo matching by modeling ground-truth distributions using dark knowledge from pre-trained networks, resulting in state-of-the-art performance on KITTI datasets and improved generalization across real-world datasets.
Despite the significant advances in domain generalized stereo matching, existing methods still exhibit domain-specific preferences when transferring from synthetic to real domains, hindering their practical applications in complex and diverse scenarios. The probability distributions predicted by the stereo network naturally encode rich similarity and uncertainty information. Inspired by this observation, we propose to extract these two types of dark knowledge from the pre-trained network to model intuitive multi-modal ground-truth distributions for both edge and non-edge regions. To mitigate the inherent domain preferences of a single network, we adopt network ensemble and further distinguish between objective and biased knowledge in the Laplace parameter space. Finally, the objective knowledge and the original disparity labels are jointly modeled as a mixture of Laplacians to provide fine-grained supervision for the stereo network training. Extensive experiments demonstrate that: (1) Our method is generic and effectively improves the generalization of existing networks. (2) PCWNet with our method achieves the state-of-the-art generalization performance on both KITTI 2015 and 2012 datasets. (3) Our method outperforms existing methods in comprehensive ranking across four popular real-world datasets.