在本次工作中,作者提出一种实时、6DoF、无需人脸检测和关键点定位的三维人脸姿态估计方法。研究中发现,估计人脸的 rigid transformation(刚体变换)比人脸关键点检测更简单,通常用于3D人脸对齐。此外,6DoF 提供了比人脸边界框标签更多的信息。 基于以上发现,作者得出以下三点贡献: 1、提出了一个容易训练的、有效的、...
Head Pose EstimationBIWIimg2poseMAE (trained with other data)3.786# 6 Compare Geodesic Error (GE)7.10# 1 Compare MAE-aligned (trained with other data)3.4# 4 Compare Geodesic Error - aligned (GE)6.23# 4 Compare Face DetectionWIDER Face (Easy)img2poseAP0.9# 24 ...
img2pose: Face Alignment and Detection via 6DoF, Face Pose Estimation Paper accepted to the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2021 Figure 1: We estimate the 6DoF rigid transformation of a 3D face (rendered in silver), aligning it with even the tiniest faces, ...
img2pose: Face Alignment and Detection via 6DoF, Face Pose Estimation V´ıtor Albiero1,∗, Xingyu Chen2,∗, Xi Yin2, Guan Pang2, Tal Hassner2 1University of Notre Dame 2Facebook AI Figure 1: We estimate the 6DoF rigid transformation of a 3D...
img2pose: Face Alignment and Detection via 6DoF, Face Pose Estimationdoi:10.1109/CVPR46437.2021.00753Vitor AlbieroXingyu ChenXi YinGuan PangTal HassnerIEEEComputer Vision and Pattern Recognition
Code to test and use the model from "Stacked Hourglass Networks for Human Pose Estimation" - princeton-vl/pose-hg-demo