CVPR Daily - Tuesday

This work aims to increase the frame rate of a video by using information from existing frames to generate intermediate ones. Given a frame pair, it seeks to pass the information from the two into the new in-between frames. Many methods have been proposed for this non-trivial task. Previous works have focused on passing information into the intermediate time step on a frame-by-frame basis to improve the quality of each frame. Ping wants to make this process easier and faster. Ping Hu is a final-year PhD student at Boston University, advised by Kate Saenko and Stan Sclaroff, working predominantly on video applications in computer vision and machine learning. In this paper, Ping proposes a new framework for video frame interpolation. He speaks to us ahead of his poster today. Many-to-Many Splatting for Efficient Video Frame Interpolation 8 DAILY CVPR Tuesday Poster Presentation