Research And Implementation Of Human Body Feature Recognition Based On Deep Learning

With the active development of sports in people's lives,more and more human body feature recognition technology is needed.This technology can assist three-dimensional human body modeling in the field of sports to analyze the movement process of people in images.At the same time,the popularity of various current shooting devices such as mobile phones and portable sports cameras has promoted the growth of image data.Therefore,through computer vision technology,image data is widely used in practical application scenarios of human body feature recognition.Moreover,in recent years,deep learning technology has made great breakthroughs in the field of computer vision,so it is possible to study the related applications of deep learning technology in human body feature recognition.In this thesis,a related method based on deep learning for human body feature recognition is proposed.The author also designs and implements a human body feature recognition system.The input of the system is a picture,and the output is the corresponding human body feature of the subject in the picture.In this thesis,the author first introduces the research background and significance,and explains the research objectives and contents.Next,the work in the related technical fields involved in the research process is introduced.Then,the needs analysis of the human body feature recognition system to be implemented is carried out to determine the functions to be completed by the system.According to the needs analysis,combined with the deep learning technology,the system outline design and function module division are explained,and the detailed design and implementation of the system is completed.Subsequently,the test cases were designed to perform white box test and black box test on the system,and the system operation effect was analyzed.Finally,the whole thesis work is summarized,and the next step of work is considered and prospected.The human body feature recognition system proposed in this thesis adopts an instance segmentation method based on deep learning,which can segment the human body contour of the subject in the picture,and identify the human body feature by the unsupervised learning method.At the same time,the system also uses the supervised learning method,combined with face detection,to predict the body BMI value of the subject in the picture.According to the test results of the system,it is verified that the human body feature recognition system can achieve the function of recognizing human body features.