Fall detection and recognition based on GCN and 2D Pose

In this paper, based on the motion properties of human skeleton, the neural network model based on graph convolution network (GCN) and 2D pose is introduced to carry out deep learning and classification of target actions, so as to detect falls. The experimental results show that our method is superior to the existing methods in two-dimensional pose, and the accuracy of the two aspects is not very different in 3D pose. However, we use GPU acceleration for testing on ubuntu platform with an average speed of 25fps, which basically meets the requirements of real-time engineering calculation and is easier to deploy to the real environment than 3D pose.