Think about boundary: Fusing multi-level boundary information for landmark heatmap regression

TL;DR

This paper introduces a boundary-aware facial landmark prediction method that leverages multi-level boundary information through a two-stage, end-to-end framework, improving accuracy especially under occlusion and pose variations.

Contribution

It proposes a novel two-module approach combining boundary estimation and landmark transformation, with multi-scale feature fusion for enhanced facial landmark prediction.

Findings

Outperforms state-of-the-art methods on benchmark datasets.

Effective boundary-aware features improve landmark localization.

Robust under occlusion and large pose variations.

Abstract

Although current face alignment algorithms have obtained pretty good performances at predicting the location of facial landmarks, huge challenges remain for faces with severe occlusion and large pose variations, etc. On the contrary, semantic location of facial boundary is more likely to be reserved and estimated on these scenes. Therefore, we study a two-stage but end-to-end approach for exploring the relationship between the facial boundary and landmarks to get boundary-aware landmark predictions, which consists of two modules: the self-calibrated boundary estimation (SCBE) module and the boundary-aware landmark transform (BALT) module. In the SCBE module, we modify the stem layers and employ intermediate supervision to help generate high-quality facial boundary heatmaps. Boundary-aware features inherited from the SCBE module are integrated into the BALT module in a multi-scale fusion framework to better model the transformation from boundary to landmark heatmap. Experimental results conducted on the challenging benchmark datasets demonstrate that our approach outperforms state-of-the-art methods in the literature.