The Development of a Smart Assistive Technology and Coaching System for Boccia - Vi Deo Analysis Implementation, Deep Learning Construction, and Field Application among Special Populations

The development trend of the intelligent vision and deep learning has raised lots of notification in commercial sports. These techniques, however, have not been widely applied to special education for the disabled yet. Individuals with disabilities usually performed not as well as typical developed individuals in motions, communication, and recognition aspects. Thus, the design of proper activities and the use of appropriate assistive devices can promote the physical development and activity participation of the disabled. Boccia is such a sport designed for people with disabilities, requiring highly coordinated motor skills and strategy planning. It has been proved that this sport can bring significant improvements in motor function, recognition ability, and social communication for the disabled. However, the game and the training of boccia participants requires much support from teachers, assistants, coaches and therapists. To promote health for the disabled and save education resources, we aim to design and develop an intelligent assistive technology and coaching system for boccia. The project includes three stages. At the first stage, we will apply the video processing technology to mark the ball positions and identify the movement features of each participant automatically. At the second stage, we will construct a deep learning model to classify the movements of each participant and analyze their skeleton parameters precisely. At the third stage, we will develop an intelligent coaching and training system which can provide different scenarios and suggestions to guide the participants with proper training strategy and practice program. The outcomes of the project will be applied to real games fields and schools for disabled community. We will also construct a cloud database to collect aa analysis results. These data can be retrieved for movement evaluation, exercise prescription, physical activity promotion, and also serve as references for more healthcare applications.

Project ID:PB10901-1066
External Project ID:MOST107-2221-E182-050-MY3