[an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive] Postdoctoral Researcher [an error occurred while processing this directive] Technical University Darmstadt, Germany [an error occurred while processing this directive] PhD '18 Swiss Federal Institute of Technology (ETH) Zurich [an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive]

Artificial Intelligence

Graphics

Computer Vision

[an error occurred while processing this directive] PhaseNet for Video Frame Interpolation [an error occurred while processing this directive] Most approaches for video frame interpolation require accurate dense correspondences to synthesize an inbetween frame. Therefore, they do not perform well in challenging scenarios with e.g. lighting changes or motion blur. Recent deep learning approaches that rely on kernels to represent motion can only alleviate these problems to some extent. In those cases, methods that use a per-pixel phase-based motion representation have been shown to work well. However, they are only applicable for a limited amount of motion. This work proposes a new approach, PhaseNet, that is designed to robustly handle challenging scenarios while also coping with larger motion compared to previous phase-based methods. Instead of working with color pixel values we use the phase-based decomposition of images as input and output to our network. The presented approach consists of a neural network decoder that directly estimates the phase decomposition of the intermediate frame. The results show that this is superior to the hand-crafted heuristics previously used in phase-based methods and also compares favorably to recent deep learning-based approaches for video frame interpolation on challenging datasets. [an error occurred while processing this directive] Simone Schaub-Meyer is a postdoctoral researcher at the Visual Inference Lab at the Technical University of Darmstadt since September 2020. Her research interests lie at the intersection of computer vision, computer graphics, and machine learning with a focus on motion representation, temporal interpolation, and video frame synthesis, with published results in top-tier vision conferences. She did her doctorate at ETH Zurich, working jointly with the Imaging and Video Processing Group at Disney Research. For her doctoral thesis, defended in 2018, she received the ETH medal, awarded yearly for outstanding theses. Prior to joining TU Darmstadt she was a postdoctoral researcher at the Media Technology Lab at ETH Zurich working on augmented reality. [an error occurred while processing this directive] Personal home page [an error occurred while processing this directive] [an error occurred while processing this directive]