Recurrent Video Restoration Transformer with Guided Deformable Attention

Benchmark Model Rank Results
analog-video-restoration-on-tapeRVRT#4PSNR: 32.47SSIM: 0.896LPIPS: 0.117VMAF: 72.41
deblurring-on-dvd-1RVRT#2PSNR: 34.92SSIM: 97.38
video-denoising-on-davis-sigma10RVRT#2PSNR: 40.57
video-denoising-on-davis-sigma20RVRT#2PSNR: 38.05
video-denoising-on-davis-sigma30RVRT#1PSNR: 36.57
video-denoising-on-davis-sigma40RVRT#1PSNR: 35.47
video-denoising-on-davis-sigma50RVRT#1PSNR: 34.57
video-denoising-on-set8-sigma10RVRT#2PSNR: 37.53
video-denoising-on-set8-sigma20RVRT#2PSNR: 34.83
video-denoising-on-set8-sigma30RVRT#2PSNR: 33.3
video-denoising-on-set8-sigma40RVRT#1PSNR: 32.21
video-denoising-on-set8-sigma50RVRT#1PSNR: 31.33
video-deraining-on-vrdsRVRT#8SSIM: 0.8857PSNR: 28.24
video-super-resolution-on-udm10-4x-upscalingRVRT#2PSNR: 40.9SSIM: 0.9729
video-super-resolution-on-vid4-4x-upscalingRVRT#7SSIM: 0.8462PSNR: 27.99
video-super-resolution-on-vid4-4x-upscaling-1RVRT#1PSNR: 29.54SSIM: 0.8810
video-super-resolution-on-vimeo90kRVRT#2PSNR: 38.59SSIM: 0.9576