Attention based Memory video portrait matting

TL;DR

This paper introduces an attention-based, trimap-free video portrait matting method that efficiently leverages temporal coherence through a novel aggregation module, reducing computational complexity compared to existing approaches.

Contribution

It presents a new attention-driven approach with a temporal aggregation module for improved video matting without requiring trimaps.

Findings

Effective temporal coherence modeling improves matting accuracy.

Reduces computational complexity compared to traditional methods.

Achieves competitive results without trimap dependency.

Abstract

We proposed a novel trimap free video matting method based on the attention mechanism. By the nature of the problem, most existing approaches use either multiple computational expansive modules or complex algorithms to exploit temporal information fully. We designed a temporal aggregation module to compute the temporal coherence between the current frame and its two previous frames.