Home

Awesome

EvUnroll: Neuromorphic Events based Rolling Shutter Image Correction (CVPR22)

This repo contains the code and dataset for EvUnroll: Neuromorphic events based rolling shutter image correction by Xinyu Zhou, Peiqi Duan, Yi Ma, and Boxin Shi.

Examples of RS correction results on Gev-RS dataset

example

Data

Gev-RS dataset follows the below directory format:

├── path_to_your_dataset_folder/
    ├── Gev-RS/
        ├── train/
            ├── seq1/
                ├── rs_blur/
                    ├── 00000.png
                    ......
                ├── rs_sharp/
                    ├── 00000.png
                    ......
                ├── gt/
                    ├── 00000.png
                    ......
            ├── seq2/
            ......
        ├── test/
            ......
    ├── all_sequence/
        ├── train/
            ├── seq1.avi
            ......
        ├── test/
            ......
    ├── Gev-RS-DVS/
        ├── train/
            ├── seq1/
                ├── seq1.h5
                ├── seq1_events_viz.avi
            ......
        ├── test/
            ......            

Usage

Dependency

pip install -r requirements.txt

Test

python test.py

Train

The training procedure consists of three steps:

To reprocude the training procedure, you might need to slightly modify the training code in different training steps, and

python train.py

License

The datasets can be freely used for research and education only. Any commercial use is strictly prohibited.

Bibtex

@InProceedings{Zhou_2022_CVPR,
    author    = {Zhou, Xinyu and Duan, Peiqi and Ma, Yi and Shi, Boxin},
    title     = {EvUnroll: Neuromorphic Events Based Rolling Shutter Image Correction},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2022},
    pages     = {17775-17784}
}

Contact

If you have any questions, please send an email to zhouxiny@pku.edu.cn