A convenient photo-based approach for assessing body posture

Good body posture is important because it helps to reduce the risk of musculoskeletal injuries and permanent distortions which may interfere with efficient functioning of the body. Individuals need to be aware that good body posture is vital for a healthy quality of life, especially as one ages. However, an individual´s ability to detect poor posture can be difficult as a faulty posture tends to feel normal after continuous practice. Moreover, consulting a physician about one´s posture may not always be feasible. As such, we propose a convenient photo-based approach for assessing body posture from a side profile image of an individual. Shape features extracted by processing these images, such as the curves from an individual´s back are used to train a Support Vector Machine and a Naïve Bayes classification models to predict good or poor posture. Results from the experiments conducted with 144 participants demonstrate the effectiveness of our approach.