Sitting posture assessment using computer vision

Abstract

Since most people usually sit most of the time nowadays because of the modern lifestyle, proper sitting posture must be exhibited so that postural abnormalities can be avoided. In order to see how proper the posture is, posture assessment is usually done by measuring the sitting parameters defining the sitting position which are the thoracic angle [TA], cervical angle [CA], retraction angle [RA], sitting height [SH], sitting eye height [SEH], sitting shoulder height [SSH], shoulder breadth [SB], hip breadth [HB], buttock-popliteal height [BPH], and the popliteal height [PH]. These sitting parameters can be measured through different methods, namely: the plumbline method, which is usually done by physical therapists; using goniometers; using accelerometers; the radiographic method; and the pressure distribution analysis through pressure sensor on a chair. However, these methods do not able to measure all the sitting parameters mentioned. Thus, there is a need to develop an algorithm that can measure and assess the parameters mentioned, which can serve as assistance for the physical therapist for posture correction. In this paper, the researchers present a method of obtaining sitting posture parameters and assess it through the use of Computer Vision in order to be used as an assistance for physical therapists in their sitting posture assessment and correction. With 42 samples, the proposed algorithm gave an accuracy of 61.9% in assessing sitting posture.