Research On Human Action Recognition Based On Deep Learning

With the progress of mankind and the development of science and technology,computer vision technology has been widely used in medical,,military,and manufacturing fields,and at the same time it has promoted the rapid development of related disciplines such as artificial intelligence and machine learning.Human body action recognition is a research topic in the field of computer vision.It has great application prospects and practical values in various fields,such as video surveillance,human-computer interaction,smart home,and sports.For the subject of human action recognition,researchers have conducted in-depth research and achieved good results.However,how to identify human movements efficiently and accurately is still a challenging problem.The human action recognition technology covers a wide range of fields,involving digital image processing,computer vision,deep learning,pattern recognition and so on.The core steps of its process are moving target detection,feature extraction and feature analysis.For feature extraction and feature analysis,the traditional recognition methods need to rely on experience to manually extract features.Feature errors and recognition effects are not ideal.The deep learning method autonomously learns image features to achieve feature extraction,which greatly simplifies the process of feature extraction.The main research work in this dissertation is to realize the recognition algorithm of human body action based on deep learning.This method automatically extracts image features through convolution kernels,which greatly improves the efficiency compared to traditional manual extraction feature methods.This method uses the deep Convolutional Neural Network model of VGG under the TensorFlow platform to perform human body action recognition.On the Matlab platform,the KTH dataset and Weizmann dataset are preprocessed to obtain the input image set.Then the preprocessed datasets are used to train the VGG model to obtain the optimal network model.Given the two datasets as test data,the corresponding confusion matrix showing the recognition result are obtained.Finally,the confusion matrix is analyzed in detail,and the specific reasons for the confusion of each action are given.At the same time,the recognition accuracy rate and the average recognition accuracy rate of each action category are counted.The experimental results show that the human action recognition algorithm based on deep learning achieves higher recognition accuracy.