Can Large Pretrained Depth Estimation Models Help With Image Dehazing?
Hongfei Zhang, Kun Zhou, Ruizheng Wu, Jiangbo Lu · 2025
Read Can Large Pretrained Depth Estimation Models Help With Image Dehazing? by Hongfei Zhang, Kun Zhou, Ruizheng Wu, Jiangbo Lu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-00698.txt.
Description
Hongfei Zhang and three colleagues present a 2025 computer vision paper asking whether depth estimators can rescue foggy photos. Haze varies across a scene, and most dehazing networks are bespoke, hard to adapt. Depth models, trained on millions of images, know scene layout well; the authors checked whether that knowledge survives haze.
It does: the deep features of a pretrained depth model stayed remarkably consistent as haze thickened. Building on that, the team designed a plug-and-play module that fuses color and depth features and slots into many dehazing architectures. Across several benchmarks the module improved results broadly, suggesting dehazing can borrow strength from large pretrained models instead of training everything from scratch.
Catalog record
| CREATOR | Hongfei Zhang, Kun Zhou, Ruizheng Wu, Jiangbo Lu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-00698