Restoring Images in Adverse Weather Conditions via Histogram Transformer


Cover figure	Network structure

Abstract

Transformer-based image restoration methods in adverse weather have achieved significant progress. Most of them use self-attention along the channel dimension or within spatially fixed-range blocks to reduce computational load. However, such a compromise results in limitations in capturing long-range spatial features. Inspired by the observation that the weather-induced degradation factors mainly cause similar occlusion and brightness, in this work, we propose an efficient Histogram Transformer (Histoformer) for restoring images affected by adverse weather. It is powered by a mechanism dubbed histogram self-attention, which sorts and segments spatial features into intensity-based bins. Self-attention is then applied across bins or within each bin to selectively focus on spatial features of dynamic range and process similar degraded pixels of the long range together. To boost histogram self-attention, we present a dynamic-range convolution enabling conventional convolution to conduct operation over similar pixels rather than neighbor pixels. We also observe that the common pixel-wise losses neglect linear association and correlation between output and ground-truth. Thus, we propose to leverage the Pearson correlation coefficient as a loss function to enforce the recovered pixels following the identical order as ground-truth. Extensive experiments demonstrate the efficacy and superiority of our proposed method.

Visual Examples


RainDrop	Outdoor-Rain	Snow100K-L	RealSnow

Installation

See INSTALL.md for the installation of dependencies required to run Histoformer.

Evaluation

cd Allweather
Download the pre-trained models and place it in ./pretrained_models/
Download test datasets from each of them, i.e., Snow100K, Outdoor-Rain, and RainDrop.
Test with the replaced argument --input_dir [INPUT_FOLDER]

python test_histoformer.py --input_dir [INPUT_FOLDER] --result_dir result/ --weights pretrained_models/net_g_best.pth --yaml_file Options/Allweather_Histoformer.yml

# for realsnow
python test_histoformer.py --input_dir [INPUT_FOLDER] --result_dir result/ --weights pretrained_models/net_g_real.pth --yaml_file Options/Allweather_Histoformer.yml

Compute PSNR and SSIM by

python compute_psnr.py --path1 [GT-PATH] --path2 [Restored-PATH]

Values may be slightly different because a) the images I upload are in JPG file format for saving space, but the values reported are computed on the PNG-format images; b) some values are reported by previous works like WeatherDiff and may be slightly different from this reproduction.

Demo

cd Allweather
Download the pre-trained models and place it in ./pretrained_models/
Test with the replaced argument --input_dir [INPUT_FOLDER]

# for realsnow
python test_histoformer.py --input_dir [INPUT_FOLDER] --result_dir result/ --weights pretrained_models/net_g_real.pth --yaml_file Options/Allweather_Histoformer.yml

Citation

If you find our work useful, please cite the following paper:

@article{sun2024restoring,
  title={Restoring Images in Adverse Weather Conditions via Histogram Transformer},
  author={Sun, Shangquan and Ren, Wenqi and Gao, Xinwei and Wang, Rui and Cao, Xiaochun},
  journal={arXiv preprint arXiv:2407.10172},
  year={2024}
}

@InProceedings{10.1007/978-3-031-72670-5_7,
    author="Sun, Shangquan and Ren, Wenqi and Gao, Xinwei and Wang, Rui and Cao, Xiaochun",
    editor="Leonardis, Ale{\v{s}} and Ricci, Elisa and Roth, Stefan and Russakovsky, Olga and Sattler, Torsten and Varol, G{\"u}l",
    title="Restoring Images in Adverse Weather Conditions via Histogram Transformer",
    booktitle="Computer Vision -- ECCV 2024",
    year="2025",
    publisher="Springer Nature Switzerland",
    address="Cham",
    pages="111--129",
    isbn="978-3-031-72670-5"
}

sunsean
/

Histoformer