Can Large Pretrained Depth Estimation Models Help With Image Dehazing?
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866918118250512384 |
|---|---|
| author | Zhang, Hongfei Zhou, Kun Wu, Ruizheng Lu, Jiangbo |
| author_facet | Zhang, Hongfei Zhou, Kun Wu, Ruizheng Lu, Jiangbo |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_00698 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Can Large Pretrained Depth Estimation Models Help With Image Dehazing? Zhang, Hongfei Zhou, Kun Wu, Ruizheng Lu, Jiangbo Computer Vision and Pattern Recognition 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. |
| title | Can Large Pretrained Depth Estimation Models Help With Image Dehazing? |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2508.00698 |