- To download DPDD training data, run
python download_data.py --data train
- Generate image patches from full-resolution training images, run
python generate_patches_dpdd.py
- To train MIRNetv2 on dual-pixel defocus deblurring task, run
cd MIRNetv2
./train.sh Defocus_Deblurring/Options/DefocusDeblur_DualPixel_16bit_MIRNet_v2.yml
Note: The above training scripts use 8 GPUs by default. To use any other number of GPUs, modify Restormer/train.sh and DefocusDeblur_DualPixel_16bit_MIRNet_v2.yml
- Download the pre-trained model and place it in
./pretrained_models/:
wget https://github.com/swz30/MIRNetv2/releases/download/v1.0.0/dual_pixel_defocus_deblurring.pth -P pretrained_models/
- Download test dataset, run
python download_data.py --data test
- Testing
python test_dual_pixel_defocus_deblur.py --save_images
This testing script will reproduce image quality scores of Table 2 in the paper.