基于多尺度编码器融合的三维人体姿态估计算法

包晓安,陈恩琳,张娜,涂小妹,吴彪,张庆琪

3D human pose estimation based on multi-scale encoder fusion

Xiaoan BAO,Enlin CHEN,Na ZHANG,Xiaomei TU,Biao WU,Qingqi ZHANG

表 2 协议1下MSEFN与不同方法在Human3.6M数据集上的MPJPE结果对比（基于GT检测姿态输入)

Tab.2 Comparison of MPJPE result of MSEFN and different methods under protocol 1 on Human3.6M dataset (based on GT-detected pose input)

GT 协议1 会议名 MPJPE/mm Dir. Disc. Eat. Grceet Phone Photo Pose Punch. Sit SitD. Somke Wait WalkD. Walk WalkT. 平均值 PoseFormer [13] (f = 81) ICCV’21 30.0 33.6 29.9 31.0 30.2 33.3 34.8 31.4 37.8 38.6 31.7 31.5 29.0 23.3 23.1 31.3 MHFormer [21] (f =351) CVPR’22 27.7 32.1 29.1 28.9 30.0 33.9 33.0 31.2 37 39.3 30.0 31.0 29.4 22.2 23.0 30.5 POT[36] (f = 81) AAAI’23 32.9 38.3 28.3 33.8 34.9 38.7 37.2 30.7 34.5 39.7 33.9 34.7 34.3 26.1 28.9 33.8 MSEFN (f =81，kf = 27) — 27.1 28.0 25.3 26.5 24.6 27.7 29.8 26.0 31.3 33.6 26.5 28.7 28.1 16.3 18.3 26.5