Advances in inertial sensing and computer vision have opened up new possibilities for obtaining accurate data in clinical and natural settings. However, the need to carefully align sensors with the body in clinical applications slows down the data collection process.

With the development of markerless motion capture, researchers have proposed a new deep learning model that utilizes data from vision, inertial sensors, and their noisy data to estimate human motion, which performs accurately in the presence of noisy data, thus avoiding the need for careful sensor-to-body calibration and a large number of cameras. And it provides data with research-grade accuracy that can help in the diagnosis, prognosis, and treatment of musculoskeletal disorders.