DynaVSR: Dynamic Adaptive Blind Video Super-Resolution

9 Nov 2020  ·  Suyoung Lee, Myungsub Choi, Kyoung Mu Lee ·

Most conventional supervised super-resolution (SR) algorithms assume that low-resolution (LR) data is obtained by downscaling high-resolution (HR) data with a fixed known kernel, but such an assumption often does not hold in real scenarios. Some recent blind SR algorithms have been proposed to estimate different downscaling kernels for each input LR image. However, they suffer from heavy computational overhead, making them infeasible for direct application to videos. In this work, we present DynaVSR, a novel meta-learning-based framework for real-world video SR that enables efficient downscaling model estimation and adaptation to the current input. Specifically, we train a multi-frame downscaling module with various types of synthetic blur kernels, which is seamlessly combined with a video SR network for input-aware adaptation. Experimental results show that DynaVSR consistently improves the performance of the state-of-the-art video SR models by a large margin, with an order of magnitude faster inference time compared to the existing blind SR approaches.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Video Super-Resolution MSU Video Super Resolution Benchmark: Detail Restoration DynaVSR-R Subjective score 6.136 # 7
ERQAv1.0 0.709 # 10
QRCRv1.0 0.557 # 10
SSIM 0.865 # 11
PSNR 28.377 # 12
FPS 0.177 # 27
1 - LPIPS 0.884 # 14
Video Super-Resolution MSU Video Super Resolution Benchmark: Detail Restoration DynaVSR-V Subjective score 4.359 # 24
ERQAv1.0 0.643 # 21
QRCRv1.0 0.549 # 14
SSIM 0.864 # 12
PSNR 29.011 # 11
FPS 0.15 # 28
1 - LPIPS 0.859 # 21
Video Super-Resolution MSU Video Upscalers: Quality Enhancement DynaVSR PSNR 26.12 # 46
SSIM 0.916 # 34
VMAF 56.86 # 7

Methods


No methods listed for this paper. Add relevant methods here