Aerobics Action Recognition Algorithm Based on Three-Dimensional Convolutional Neural Network and Multilabel Classification

In the context of modern people increasingly paying attention to health and promoting aerobics, amount data audiences aerobics videos has grown rapidly, its potential application value attracted widespread from scientific research industry perspectives. This article integrated computer vision deep learning related knowledge realize intelligent recognition representation specific human movements in video sequences. The study proposes an automatic method for floor exercise based on three-dimensional convolutional networks multilabel classification. Since two-dimensional neural (CNNs) lose time information when extracting features, so overcome this, proposed uses perform recognition. feature is taken space, extracted features are subjected multiple binary classifications achieve goal Various comparison simulation experiments conducted research, experimental results prove effectiveness superiority approach.