Early Childhood Gymnastic Motion Recognition System Using Image Processing Technology
RS Asmaul Husna, Andani Achmad, Amil Ahmad Ilham, Zahir Zainuddin, Arsan Kumala Jaya · 2020
It is very important to assess the motor development of students in Early Childhood Education (PAUD) schools. The motor skills of students can be evaluated through the ability of children to follow the gymnastic movements taught by the teacher. A large number of students makes it difficult to monitor the development of children's movements directly. In this study, it is proposed to monitor the development of students through video recordings of learning exercises in the classroom. Learning videos of students' gymnastics are turned into digital images. The object (students) on the frame are recognized using the Histogram of Oriented Gradient (HOG) method and the student's gymnastic movements are detected using Principal Component Analysis (PCA). The experiment used 280 pictures as training data, the training data consisted of 8 students' gymnastic movements, each movement used 35 training data. In the testing phase using 4 video input learning exercises of students' gymnastics, 1 video input demonstrates 8 gymnastic movements with a total of 4 students. The experimental results show an accuracy rate of 96,09%.