Automatic Portrait Video Matting via Context Motion Network

TL;DR

This paper introduces a context motion network for automatic portrait video matting that effectively combines semantic and motion information, leveraging optical flow and recurrent feature integration to improve performance over existing methods.

Contribution

The paper proposes a novel context motion network that incorporates optical flow and a recurrent feature updating mechanism for enhanced portrait video matting.

Findings

Outperforms state-of-the-art methods on Video240K SD dataset

Effectively leverages temporal motion information for better matting

Demonstrates significant performance improvements over existing approaches

Abstract

Automatic portrait video matting is an under-constrained problem. Most state-of-the-art methods only exploit the semantic information and process each frame individually. Their performance is compromised due to the lack of temporal information between the frames. To solve this problem, we propose the context motion network to leverage semantic information and motion information. To capture the motion information, we estimate the optical flow and design a context-motion updating operator to integrate features between frames recurrently. Our experiments show that our network outperforms state-of-the-art matting methods significantly on the Video240K SD dataset.