Shuffle Transformer with Feature Alignment for Video Face Parsing

TL;DR

This paper presents a Shuffle Transformer backbone combined with a Feature Alignment Aggregation module to improve face parsing accuracy in short videos, achieving first place in a CVPR challenge.

Contribution

The introduction of a cross-window Shuffle Transformer backbone and FAA module for enhanced face parsing accuracy and edge detail preservation.

Findings

Achieved 86.95% score in the PIC workshop challenge.

Ranked first in the Short-video Face Parsing track.

Demonstrated improved segmentation quality with the proposed method.

Abstract

This is a short technical report introducing the solution of the Team TCParser for Short-video Face Parsing Track of The 3rd Person in Context (PIC) Workshop and Challenge at CVPR 2021. In this paper, we introduce a strong backbone which is cross-window based Shuffle Transformer for presenting accurate face parsing representation. To further obtain the finer segmentation results, especially on the edges, we introduce a Feature Alignment Aggregation (FAA) module. It can effectively relieve the feature misalignment issue caused by multi-resolution feature aggregation. Benefiting from the stronger backbone and better feature aggregation, the proposed method achieves 86.9519% score in the Short-video Face Parsing track of the 3rd Person in Context (PIC) Workshop and Challenge, ranked the first place.