Video frame interpolation via down–up scale generative adversarial networks

作者：

Highlights：

• Introduce an improved multi-scale generative adversarial network for frame interpolation.

• Propose a two-scale generator in the generative adversarial network, where the down-scaled-input module captures the overall structure of the scene while the original-scale-input module restores finer textures.

• Achieve the most satisfactory trade-off between the synthesis quality and runtime compared to other state-of-the-art frame interpolation approaches.

• Illustrate comprehensive experiments to show the superior performance of the proposed framework.

摘要

•Introduce an improved multi-scale generative adversarial network for frame interpolation.•Propose a two-scale generator in the generative adversarial network, where the down-scaled-input module captures the overall structure of the scene while the original-scale-input module restores finer textures.•Achieve the most satisfactory trade-off between the synthesis quality and runtime compared to other state-of-the-art frame interpolation approaches.•Illustrate comprehensive experiments to show the superior performance of the proposed framework.

论文关键词：Video frame interpolation,Video frame generation,Deep learning,Generative adversarial networks,Generation network

论文评审过程：Received 8 July 2021, Revised 3 April 2022, Accepted 13 April 2022, Available online 26 April 2022, Version of Record 7 May 2022.

论文官网地址：https://doi.org/10.1016/j.cviu.2022.103434