Video Enhancement with Task-Oriented Flow
This work addresses the need for better motion representations in video processing for applications like video editing and restoration, though it is incremental as it builds on existing flow-based methods.
The paper tackles the problem of sub-optimal optical flow for video enhancement by proposing task-oriented flow (TOFlow), a self-supervised motion representation learned jointly with video processing tasks, which outperforms traditional optical flow on benchmarks and a new dataset in frame interpolation, denoising/deblocking, and super-resolution.
Many video enhancement algorithms rely on optical flow to register frames in a video sequence. Precise flow estimation is however intractable; and optical flow itself is often a sub-optimal representation for particular video processing tasks. In this paper, we propose task-oriented flow (TOFlow), a motion representation learned in a self-supervised, task-specific manner. We design a neural network with a trainable motion estimation component and a video processing component, and train them jointly to learn the task-oriented flow. For evaluation, we build Vimeo-90K, a large-scale, high-quality video dataset for low-level video processing. TOFlow outperforms traditional optical flow on standard benchmarks as well as our Vimeo-90K dataset in three video processing tasks: frame interpolation, video denoising/deblocking, and video super-resolution.