Supervised Fog Removal Dataset
Overview
This dataset contains 80,000 paired images designed for supervised image dehazing / fog removal tasks.
Each sample consists of:
a clean image (ground truth)
a synthetically fogged version of that image
The fog is generated using a physics-inspired atmospheric scattering model combined with depth estimation, allowing the fog to behave realistically with respect to scene geometry.
Unlike simple uniform haze overlays, this dataset simulates depth-aware fog… See the full description on the dataset page:
https://huggingface.co/datasets/Aeye-coder/Supervised-Fog-Removal-Dataset.
