Similarity index retrieval for school kids' sitting posture identification

A good sitting posture is very important in one's individual life. Since the past time, it has been said that a person who sit in a good posture is more able to focus on their studies. In today's world, a lot of students and adults are suffering from this problem of poor sitting posture due to which the number of people at a physiotherapist's center is increasing. These physiotherapists help them to correct their posture. The main objective of developing the sitting posture is that it will help the students to correct their posture since school time, which can help them in future to sit in a good posture. A good sitting posture enhances blood flow and safeguards the health of the nerves and blood vessels. This paper describes the standing upright and in the proper alignment of a person. Therefore, it is highly useful to accurately monitor students' sitting posture in the classroom and alert them of poor posture. This chapter suggests a sitting posture for students during class by using a variety of student images as raw data for a model that will scan the images and determine whether the student is seated correctly. The model will use OpenPose, which provides the posture feature, to determine whether the student is hunched over, folding their hands, kneeling, and sitting straight. Convolutional neural networks are built using Keras datasets to identify student the sitting posture. This model will ultimately state the percentage error of a speciiňAc image that it will scan in a speciiňAc amount of time. The accuracy of the model that we built getting is 98% at the time when there is nothing behind the object.