Vision-Based System for Driver Posture Tracking to Prevent Musculoskeletal Disorders

This paper investigates the use of OpenPose and Kinect to monitor the driver’s upper body posture with the aim to detect unhealthy postures on the road that can lead to musculoskeletal disorders. A comparison between the two methods revealed that although the former has been highly used in human motion tracking, it has reached its limits and was discontinued. The three models of the novel open-source OpenPose were evaluated and Body_25 configuration was found to offer the best results in joints detection and tracking. A preliminary assessment with ten participants indicated that OpenPose and Kinect can successfully be used in monitoring the driver’s sitting posture. A subjective evaluation showed that participants did not notice a difference between the two scenarios and believe that posture monitoring systems should be integrated in vehicles to increase driver safety.