JCSE, vol. 17, no. 1, pp.13-19, 2023

DOI: http://dx.doi.org/10.5626/JCSE.2023.17.1.13

Knowledge Distillation for Optical Flow-Based Video Superresolution

Jungwon Lee and Sang-hyo Park
School of Computer Science and Engineering, Kyungpook National University, Daegu, Korea

Abstract: Recently, deep learning-based super-resolution (SR) models have been used to improve SR performance by equipping preprocessing networks with baseline SR networks. In particular, in video SR, which creates a high-resolution (HR) image with multiple frames, optical flow extraction is accompanied by a preprocessing process. These preprocessing networks work effectively in terms of quality, but at the cost of increased network parameters, which increase the computational complexity and memory consumption for SR tasks with restricted resources. One well-known approach is the knowledge distillation (KD) method, which can transfer the original model???knowledge to a lightweight model with less performance degradation. Moreover, KD may improve SR quality with reduced model parameters. In this study, we propose an effective KD method that can effectively reduce the original SR model parameters and even improve network performance. The experimental results demonstrated that our method achieved a better PSNR than the original state-ofthe-art SR network despite having fewer parameters.

Keyword: Video super-resolution; Optical flow; Knowledge distillation; Deep learning; Super-resolution

Full Paper:   118 Downloads, 836 View