Can Large Pretrained Depth Estimation Models Help With Image Dehazing?
This work addresses image dehazing for computer vision applications, but it is incremental as it builds on existing pretrained models with a new fusion module.
The paper tackled the problem of image dehazing by investigating the generalization of pretrained depth representations, finding that deep depth features are consistent across haze levels, and proposed a plug-and-play RGB-D fusion module that integrates with various architectures, achieving validated effectiveness across benchmarks.
Image dehazing remains a challenging problem due to the spatially varying nature of haze in real-world scenes. While existing methods have demonstrated the promise of large-scale pretrained models for image dehazing, their architecture-specific designs hinder adaptability across diverse scenarios with different accuracy and efficiency requirements. In this work, we systematically investigate the generalization capability of pretrained depth representations-learned from millions of diverse images-for image dehazing. Our empirical analysis reveals that the learned deep depth features maintain remarkable consistency across varying haze levels. Building on this insight, we propose a plug-and-play RGB-D fusion module that seamlessly integrates with diverse dehazing architectures. Extensive experiments across multiple benchmarks validate both the effectiveness and broad applicability of our approach.