QUANTITATIVE EVALUATION OF YOGA POSTURES USING IMAGE PROCESSING AND HOTELLING’S T² STATISTICAL ANALYSIS

The aim of this study is to use computer vision and statistical analysis to measure how yoga practice can improve body posture. A Python-based model was developed that can recognize different yoga poses from images and then create a 3D skeleton of the human body using landmark points. For each pose, twelve important landmarks such as shoulders, elbows, hips, and knees were identified, and angles were calculated to check the correctness of posture. To evaluate whether these landmarks showed improvement after one month of yoga practice, we applied Hotelling’s T² test, a multivariate statistical method that can detect overall changes across several joints at the same time. The results showed that some landmarks had significant differences before and after yoga, meaning that the posture became more aligned and balanced. This method provides an objective way of checking yoga progress instead of relying only on visual observation. The study demonstrates that by combining image processing with statistical testing, it is possible to give meaningful feedback to yoga practitioners, trainers, and even rehabilitation experts in a simple and scientific manner.