A Sports Training Video Classification Model Based on Deep Learning

A sports training video classification model based on deep learning is studied for targeting low accuracy caused by the randomness of objective movement in video. The camera calibration technology used to restore position target real three-dimensional space. After video, preprocessed. input segment divided into equal length segments obtain subvideo segment. motion vector field, brightness feature, color and texture feature are extracted, extracted features AlexNet convolutional neural network. ReLU as activation function this Local response normalization suppress enhance output neurons highlight performance useful information, so that results more accurate. Event matching method match network complete classification. experimental proposed study show can effectively solve problems moving randomness. than 99%, speed faster which shown from experiments.