Data-Free Point Cloud Network for 3D Face Recognition

Abstract

Point clouds-based Networks have achieved great attention in 3D object classification, segmentation and indoor scene semantic parsing. In terms of face recognition, 3D face recognition method which directly consume point clouds as input is still under study. Two main factors account for this: One is how to get discriminative face representations from 3D point clouds using deep network; the other is the lack of large 3D training dataset. To address these problems, a data-free 3D face recognition method is proposed only using synthesized unreal data from statistical 3D Morphable Model to train a deep point cloud network. To ease the inconsistent distribution between model data and real faces, different point sampling methods are used in train and test phase. In this paper, we propose a curvatureaware point sampling(CPS) strategy replacing the original furthest point sampling(FPS) to hierarchically downsample feature-sensitive points which are crucial to pass and aggregate features deeply. A PointNet++ like Network is used to extract face features directly from point clouds. The experimental results show that the network trained on generated data generalizes well for real 3D faces. Fine tuning on a small part of FRGCv2.0 and Bosphorus, which include real faces in different poses and expressions, further improves recognition accuracy. * Corresponding Author tremely rely on face alignment algorithm and feature descriptors, which limit the scalability as compared to the endto-end learning approaches.

3D face recognition: A comprehensive survey in 2022

“…The easiest is to rotate and crop existing face data. Another popular approach is to use 3D morphable facial models (3DMM) [50] to generate new shapes and expressions to synthesize new facial data [51,52]. Randomly selecting sub-feature sets from different samples of a person and combining them to generate a new face is also a reliable way to enrich the identities in datasets [42].…”

Combining Color Texture and Shape Features In a Multi-Input Convolutional Neural Network for Efficient Face Recognition In Unconstrained Environments

“…The experiments showed that this method achieves very competitive face recognition performance. Authors in [18] proposed a data-free method for 3D face recognition using generated data from Gaussian Process Morphable Models (GPMM). Authors in [19] used the meshDOG as a detector to capture the local information of the face surface.…”

Accumulating weighted segmentation in 3D face recognition

“…Zheng et al (2019) proposed an end-to-end face recognition system combining the geometric invariants, histogram of oriented gradients and the fine-tuned ResNet. Zhang et al (2019) used a PointNet++ like network to extract face feature directly from point clouds. These methods usually require the design of a complex Deep Network or CNN, and also require laborious fine tune to obtain satisfying experimental results.…”

“…Theoretically, poor data/image quality in training data set such as occlusion/data loss will lower the performance of their method. Other approaches based on deep network/deep learn such asZhang et al (2019);Zheng et al (2019) andGilani et al (2018) are very popular in recently. Thanks to the advantages of deep network technique, these methods achieve promising results in face recognition.…”