By Jinshan Pan, Boming Xu, Haoran Bai, Jinhui Tang, and Ming-Hsuan Yang

　　　　　　
　　　　　　　　　　　　　GOPRO　　　　　　　　　　　　　　　　　　　DVD

　　　　　　
　　　　　　　　　　　　　　　　　　　　　　　BSD(1ms8ms)

This repository is the official PyTorch implementation of our TPAMI paper"Cascaded Deep Video Deblurring Using Temporal Sharpness Prior and Non-local Spatial-Temporal Similarity"

[2022-04-30] Paper has been submitted to IEEE TPAMI
[2023-01-23] Paper has been accepted by IEEE TPAMI
[2023-01-31] Training & Testing code is available!

Deblurred result on a real challenging video. Our method is motivated by the success of deblurring approaches based on variational models. It explores sharp pixels from adjacent frames by a temporal sharpness prior and non-local spatial-temporal similarity contexts to constrain deep convolutional neural networks (CNNs) and restores sharp videos by a cascaded inference process. We show that enforcing the temporal sharpness prior and non-local spatial-temporal similarity contexts and learning the deep CNNs by a cascaded inference manner can make the deep CNN more compact and generate better-deblurred results (i.e., (f) and (g)) than both the CNN-based methods [2], [3], [4] and variational model-based method [1].

Quantitative evaluations on the DVD dataset in terms of PSNR and SSIM. * denotes the reported results using self-ensemble strategy.

Quantitative evaluations on the GoPro dataset in terms of PSNR and SSIM. * denotes the reported results using self-ensemble strategy

Quantitative evaluations on the BSD video deblurring dataset in terms of PSNR and SSIM.

• Linux (Tested on Ubuntu 18.04)
• Python 3 (Recommend to use Anaconda)
• PyTorch 1.8.0: conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=10.2 -c pytorch
• Install dependent packages :pip install -r requirements.txt
• Install CDVD-TSPNL :python setup.py develop

• Models are available in './experiments/pretrained_models/'

If you prepare your own dataset, please follow the following form like GOPRO/DVD:

|--dataset  
    |--blur  
        |--video 1
            |--frame 1
            |--frame 2
                ：  
        |--video 2
            :
        |--video n
    |--gt
        |--video 1
            |--frame 1
            |--frame 2
                ：  
        |--video 2
        	:
        |--video n

• FlowNet pretrained model has been downloaded in './pretrained_models/flownet/'
• Download training dataset like above form.
• Run the following commands:

Single GPU
python basicsr/train.py -opt options/train/Deblur/train_Deblur_GOPRO.yml
Multi-GPUs
python -m torch.distributed.launch --nproc_per_node=8 --master_port=4321 basicsr/train.py -opt options/train/Deblur/train_Deblur_GOPRO.yml --launcher pytorch

• Models are available in './experiments/pretrained_models/'.
• Organize your dataset(GOPRO/DVD/BSD) like the above form.
• Run the following commands:

python basicsr/test.py -opt options/test/Deblur/test_Deblur_GOPRO.yml
cd results
python merge_full.py

• Before running merge_full.py, you should change the parameters in this file of Line 5,6,7,8.
• The deblured result will be in './results/dataset_name/'.
• If you set flip_seq: Ture in config files, testing code will use self-ensemble strategy.(CDVDTSPNL+)
• We calculate PSNRs/SSIMs following [Here]

@InProceedings{Pan_2023_TPAMI,
    author = {Pan, Jinshan and Xu, Boming and Bai, Haoran and Tang, Jinhui and Yang, Ming-Hsuan},
    title = {Cascaded Deep Video Deblurring Using Temporal Sharpness Prior and Non-local Spatial-Temporal Similarity},
    booktitle = {IEEE Transactions on Pattern Analysis and Machine Intelligence(TPAMI)},
    month = {Jan},
    year = {2023}
}