Dual Attention MobDenseNet(DAMDNet) for Robust 3D Face Alignment

TL;DR

This paper introduces DAMDNet, a dual attention-based end-to-end framework for robust 3D face alignment from monocular images, improving accuracy and efficiency especially for disguised faces.

Contribution

The paper proposes a novel dual attention mechanism combined with lightweight modules and data augmentation for enhanced 3D face alignment and reconstruction.

Findings

Significant accuracy improvement on AFLW and AFLW2000-3D datasets.

Effective handling of disguised faces with high robustness.

Reduced model complexity and parameters.

Abstract

3D face alignment of monocular images is a crucial process in the recognition of faces with disguise.3D face reconstruction facilitated by alignment can restore the face structure which is helpful in detcting disguise interference.This paper proposes a dual attention mechanism and an efficient end-to-end 3D face alignment framework.We build a stable network model through Depthwise Separable Convolution, Densely Connected Convolutional and Lightweight Channel Attention Mechanism. In order to enhance the ability of the network model to extract the spatial features of the face region, we adopt Spatial Group-wise Feature enhancement module to improve the representation ability of the network. Different loss functions are applied jointly to constrain the 3D parameters of a 3D Morphable Model (3DMM) and its 3D vertices. We use a variety of data enhancement methods and generate large virtual pose face data sets to solve the data imbalance problem. The experiments on the challenging AFLW,AFLW2000-3D datasets show that our algorithm significantly improves the accuracy of 3D face alignment. Our experiments using the field DFW dataset show that DAMDNet exhibits excellent performance in the 3D alignment and reconstruction of challenging disguised faces.The model parameters and the complexity of the proposed method are also reduced significantly.The code is publicly available at https:// github.com/LeiJiangJNU/DAMDNet