Samenvatting

About 15 million suffer from strokes annually. Strokes are caused by cardiovascular problems, leading to paralysis or weakness of the extremities affecting patient independence thus requiring rehabilitation. Recently, virtual reality and other gaming systems have been developed and tested for functional rehabilitation and especially home use. This study examines the suitability of the Microsoft Kinect v2 in detecting Activities of Daily Living (ADL) within a hand function assessment procedure called the Southampton Hand Assessment Procedure (SHAP) using the adaptive boosting algorithm within the Kinect v2 SDK. Data from a group of 14 students was used. The results showed that the ADL tasks comprising; pouring from a carton, pouring from a jug, opening a glass jar, picking up coins, lifting a light tin and heavy jar across a barrier were detected with Root Mean Square confidence values below the minimum 0.95 recommended for deployment in an application or game. This implies further data would be required to train the detector for reliable detection. The use of mixed grasping strategies for some tasks may have affected the initial detection accuracy for the last two detectors. There is potential of the Kinect v2 in stroke rehabilitation and assessment.