基于SMPL模态分解与嵌入融合的多模态步态识别

吴越,梁铮,高巍,杨茂达,赵培森,邓红霞,常媛媛

Multi-modal gait recognition based on SMPL model decomposition and embedding fusion

Yue WU,Zheng LIANG,Wei GAO,Maoda YANG,Peisen ZHAO,Hongxia DENG,Yuanyuan CHANG

表 5 多模态结构、自适应帧关节注意力模块及模态嵌入融合模块的消融实验

Tab.5 Ablation study on multimodal structure, AFJAtt, and MEFusion

轮廓分支 SMPL分支 AFJAtt MEFusion R-1 R-5 mAP/% mINP/% EFusion MCLoss — √ — — — 26.40 41.90 17.44 10.23 √ — — — — 64.90 82.20 54.96 35.70 √ √ — — — 66.10 83.20 55.64 36.44 √ √ √ — — 68.90 84.20 58.94 39.11 √ √ √ √ — 69.50 85.10 60.61 41.22 √ √ √ √ √ 70.40 85.00 61.04 41.27