FP-Net:基于任意角度单幅人体图像的正面姿态估计  

FP-Net:Frontal Pose Estimation Based on Single Human Body Image with Arbitrary View

作　　者：陈路飞 张勇 唐永正 尹宝才 Chen Lufei;Zhang Yong;Tang Yongzheng;Yin Baocai(Beijing Artificial Intelligence Institute,Faculty of Information Technology,Beijing University of Technology,Beijing 100124)

机构地区：[1]北京工业大学信息学部北京人工智能研究院,北京100124

出　　处：《计算机辅助设计与图形学学报》2022年第10期1604-1612,共9页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(62072015,U19B2039);北京市自然科学基金(4172003);中国高等教育学会专项课题(2020XXHYB16).

摘　　要：准确提取人体正面姿态,能够更好地辅助进行行为识别、图像生成和虚拟试衣工作.然而,侧面图像、背面图像存在人体自遮挡、关键点不可见等问题,使正面姿态提取非常困难.因此,设计并实现了基于任意角度单幅人体图像的正面姿态估计网络FP-Net(frontal pose network).首先,制作了多角度人体图像数据集,为模型设计提供数据支持;其次,为了提高模型预测结果的准确性,设计了基于Anchor pose的回归模块和基于三维姿态的特征融合模块;最后,通过FP-Net实现了任意角度人体图像的正面姿态提取.在BJUT Taichi和CMU Panoptic数据集上采用PCK评价指标进行消融实验,结果表明所提方法有效地提高了人体正面姿态估计的准确率.Accurate extraction of the frontal pose of the human body can assist behavior recognition,image generation,and virtually try-on work.However,side images and back images have problems,such as self-occlusion of the human body and invisible key points,which make the extraction of the frontal pose very difficult.Therefore,a frontal pose estimation network FP-Net(frontal pose network)based on single human body image at arbitrary view is designed and implemented.First,a multi-view human body image data set is produced to provide data support for model design.Second,in order to improve the accuracy of model prediction results,a regression module based on Anchor pose and a feature fusion module based on 3D pose are designed.Finally,FP-Net realizes the frontal pose extraction of human body images from arbitrary view.The PCK evaluation index is used for ablation experiments on the BJUT Taichi and CMU Panoptic datasets.The results show that proposed method effectively improves the accuracy of the frontal pose estimation of the human body.

关 键 词：姿态估计 行为分析 多角度图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]