Fusing Spatiotemporal Features and Joints for 3D Action Recognition | IEEE Conference Publication | IEEE Xplore