Learning directly from synthetic point clouds for “in-the-wild” 3D face recognition

作者：

Highlights：

• Using deep neural network directly input point clouds for 3D face recognition.

• Using learn from synthetic method based on Gaussian Process Morphable Model (GPMM) to synthesize large-scale training data.

• Curvature-aware sampling strategy together with a novel transfer learning method to learn 3D discriminative features.

摘要

•Using deep neural network directly input point clouds for 3D face recognition.•Using learn from synthetic method based on Gaussian Process Morphable Model (GPMM) to synthesize large-scale training data.•Curvature-aware sampling strategy together with a novel transfer learning method to learn 3D discriminative features.

论文关键词：3D face recognition,Learning from synthetic,Curvature-aware point sampling,Transfer learning

论文评审过程：Received 2 March 2021, Revised 11 October 2021, Accepted 21 October 2021, Available online 23 October 2021, Version of Record 30 October 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108394