DeblurGAN-v2: Deblurring (Orders-of-Magnitude) Faster and Better

We present a new end-to-end generative adversarial network (GAN) for single image motion deblurring, named DeblurGAN-v2, which considerably boosts state-of-the-art deblurring efficiency, quality, and flexibility. DeblurGAN-v2 is based on a relativistic conditional GAN with a double-scale discriminator... For the first time, we introduce the Feature Pyramid Network into deblurring, as a core building block in the generator of DeblurGAN-v2. It can flexibly work with a wide range of backbones, to navigate the balance between performance and efficiency. The plug-in of sophisticated backbones (e.g., Inception-ResNet-v2) can lead to solid state-of-the-art deblurring. Meanwhile, with light-weight backbones (e.g., MobileNet and its variants), DeblurGAN-v2 reaches 10-100 times faster than the nearest competitors, while maintaining close to state-of-the-art results, implying the option of real-time video deblurring. We demonstrate that DeblurGAN-v2 obtains very competitive performance on several popular benchmarks, in terms of deblurring quality (both objective and subjective), as well as efficiency. Besides, we show the architecture to be effective for general image restoration tasks too. Our codes, models and data are available at: https://github.com/KupynOrest/DeblurGANv2 read more

PDF Abstract ICCV 2019 PDF ICCV 2019 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Deblurring GoPro DeblurGANv2-MobileNet PSNR 28.17 # 23
SSIM 0.925 # 18
Deblurring GoPro DeblurGANv2-MobileNet-DSC PSNR 28.03 # 24
SSIM 0.922 # 19
Deblurring GoPro DeblurGANv2-Inception PSNR 29.55 # 19
SSIM 0.934 # 15
Deblurring HIDE (trained on GOPRO) DeblurGAN-v2 PSNR (sRGB) 26.61 # 8
SSIM (sRGB) 0.875 # 8
Deblurring RealBlur-J DeblurGAN-v2 SSIM (sRGB) 0.870 # 4
PSNR (sRGB) 29.69 # 4
Deblurring RealBlur-J (trained on GoPro) DeblurGAN-v2 PSNR (sRGB) 28.70 # 1
SSIM (sRGB) 0.866 # 3
Deblurring RealBlur-R DeblurGAN-v2 PSNR (sRGB) 36.44 # 3
SSIM (sRGB) 0.935 # 3
Deblurring RealBlur-R (trained on GoPro) DeblurGAN-v2 PSNR (sRGB) 35.26 # 5
SSIM (sRGB) 0.944 # 5

Results from Other Papers


Task Dataset Model Metric Name Metric Value Rank Uses Extra
Training Data
Source Paper Compare
Blind Face Restoration CelebA-Test DeblurGANv2* LPIPS 40.01 # 2
FID 52.69 # 3
NIQE 4.917 # 3
Deg. 39.64 # 2
PSNR 25.91 # 1
SSIM 0.6952 # 1

Methods