PBFG: A New Physically-Based Dataset and Removal of Lens Flares and Glares (ICCV 2025)
_{Official Implementation}

This repository contains the official dateset and code in the following paper:

PBFG: A New Physically-Based Dataset and Removal of Lens Flares and Glares.

Jie Zhu¹, Sungkil Lee¹

¹Sungkyunkwan University

International Conference on Computer Vision （ICCV） 2025

Flare and glare are common nighttime artifacts that degrade image quality and hinder computer vision tasks. Existing synthetic datasets lack physical realism and diversity, while deep learning-based removal methods struggle in complex scenes, posing significant challenges. To address these issues, we introduce the high-quality annotated Physically-Based Flare and Glare (PBFG) dataset and a Flare and Glare Removal Network (FGRNet). PBFG comprises 2,600 flares and 4,000 glares using our computational rendering scheme with diverse lens systems and optical configurations. Our advanced streak synthesis enhances template fidelity and improves streak removal accuracy. FGRNet leverages spatial-frequency features for comprehensive local and global feature extraction. It introduces a Spatial-Frequency Enhanced Module with a Spatial Reconstruction Unit and a Frequency-Enhanced Unit to extract multi-scale spatial information and enhance frequency representation. This design effectively removes complex artifacts, including large-area glares, diverse flares, and multiple or off-screen-induced streaks. Additionally, a histogram-matching module ensures stylistic and visual consistency with ground truth. Extensive experiments confirm that PBFG accurately replicates real-world patterns, and FGRNet outperforms state-of-the-art methods both quantitatively and qualitatively, resulting in significant gains of PSNRs (up to 2.3 dB and 3.14 dB in an image and its glare regions, respectively).

We use Git Large File Storage (LFS) to manage the PBFG dataset. So please use LFS to download the dataset.

The PBFG Dataset is in dataset/.

You can also download the dataset from Google Drive.

The PBStar Dataset is in dataset/PBStar.zip.

You can also download the dataset from Google Drive.

1. Clone the repo

   git clone https://github.com/cgskku/pbfg.git

2. Install dependent packages

   cd pbfg
   pip install -r requirements.txt

3. Install pbfg
   Please run the following commands in the pbfg root path to install pbfg:
   

   python setup.py develop

Our pretrained checkpoint is in code/PBFG/experiments/checkpoint.zip.

The test dataset is in dataset/test_data.zip.

To estimate the flare/glare-free images with our checkpoint pretrained on PBFG, you can run the test.py by using:

python test.py --gt dataset/test_data/real/gt  --input dataset/test_data/real/input   --output result/test_real/pbfg/  --model_path experiments/checkpoint/net_g_last.pth   --flare7kpp

To calculate different metrics with our pretrained model, you can run the evaluate.py by using:

python evaluate.py --input result/test_real/pbfg/blend/  --gt  dataset/test_data/real/gt/ --mask  dataset/test_data/real/mask/

Training with single GPU

To train a model with your own data/model, you can edit the options/uformer_flare7kpp_baseline_option.yml and run the following codes. You can also add --debug command to start the debug mode:

python basicsr/train.py -opt options/uformer_flare7kpp_baseline_option.yml

Training with multiple GPU

You can run the following command for the multiple GPU tranining:

CUDA_VISIBLE_DEVICES=0,1 bash scripts/dist_train.sh 2 options/uformer_flare7kpp_baseline_option.yml

├── PBFG
    ├── flare
    ├── glare
         ├── compound_glare
         ├── glow
         ├── light_source
         ├── shimmer
         ├── starburst
         ├── streak
├── PBStar

This project is licensed under CC BY-NC-SA 4.0. Redistribution and use of the dataset and code for non-commercial purposes should follow this license.

If you find this work useful, please cite:

@InProceedings{Zhu_2025_ICCV,
    author    = {Zhu, Jie and Lee, Sungkil},
    title     = {PBFG: A New Physically-Based Dataset and Removal of Lens Flares and Glares},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2025},
    pages     = {5448-5457}
}

This repository is based on the Flare7K. Thanks for their awesome work.