查看文章

Analysis of human posture has many applications in the field of sports and medical science including patient monitoring, lifestyle analysis, elderly care etc. It is important to understand if a person is healthy (in terms of his everyday posture) or is suffering from a joint/bone disease as reflected by his incorrect posture. Many of the works in this area have been based on computer vision techniques. These are limited in providing real-time solution. The aim of the proposed work is to classify the human posture during three different activities (standing, sitting and sleeping/lying) as a healthy or an unhealthy one. This is done by applying machine learning techniques on a large posture dataset which is collected with the help of MPU-6050 sensors mounted on multiple positions on the body. The performance evaluation of the proposed work reveals that the proposed work is efficient enough to classify the postures accurately.