Background and goals
The study of human motion within sports and rehabilitation has made significant progress during the last decades. Using computer vision (CV) based techniques; applications for realtime and automatic analysis of human body movement can be developed. These applications can range from automatic rehabilitation aids giving real time feedback and instruction to clients performing physiotherapeutic exercises to software analyzing the technical performance of a professional athlete. The advantage of using CV is that the only technical equipment needed is a camera/cameras attached to a computing device performing the motion analysis. Traditionally CVbased solutions for motion analysis though involved installation of markers (such as white dots or other reflective material) on different parts of the body. A significantly more practical and easy-to- use solution is a CV based marker-less motion detection system, though collecting accurate motion data without the use of markers is a challenge and a relevant research topic.
The goal is to gain knowledge and deep understanding on how effective algorithms for joint detection and motion analysis can be trained and improved to be able to not only perform motion analysis in a 2D but also in a 3D space. This would provide the possibility to perform more advanced motion analysis such as shoulder and hip movements. The accuracy of marker-less CV based motion analysis in both 2D and 3D space will be evaluated and compared to motion analysis performed by human physiotherapists.
Objectives and benefits
The long-term objective is to in close collaboration with students, companies, and partner universities develop prototype applications that can be register as health technology and provide automatic guidance for clients doing rehabilitation exercises by using a low-cost and easy to use device, such as a computer with a web camera or a mobile device.
A review paper on the potential of computer vision based marker-less motion analysis for rehabilitation has been published in the journal of Rehabilitation Process and Outcome in July 2021.
Prototype applications for CV-based marker-less knee angle measurements have been developed and tested by information technology and physiotherapy students. The development process from a pedagogical point of view has been presented in an Arcada working paper in March 2021.
The project has resulted in degree theses for both information technology and physiotherapy bachelor students.
The project contributes to telerehabilitation and motion analysis applications for preemptive health care which is important for (among others) the aging population.
The aging population and the recent corona pandemic, among others, have increased the need for easy-to-use, cost effective and reliable telerehabilitation services. Computer vision-based marker-less human pose estimation is a promising technique for telerehabilitation as the only technical equipment needed is a camera and a computing device. With this equipment, the rehabilitation application is able to analyse and supervise clients’ exercises and reduce clients’ need for visiting physiotherapists in person.The long-term goal is to in close collaboration with students, companies, and partner universities develop prototype applications that can be register as health technology and provide automatic guidance for clients doing rehabilitation exercises.