Classification of Stroke and Non-Stroke Patients from Human Body Movements using Smartphone Videos and Deep Neural Networks

Abstract

This study covers a pilot study on developing a tele-health system for detection and classification of stroke and non-stroke patients from human body movements using smartphone videos. Human body poses are extracted from smartphone videos which are then transformed into RGB images and classified into either stroke (positive) or non-stroke (negative) labels. We tested PoseNet, BlazePose, and MoveNet for human body pose detection and AlexN et and SqueezeN et for classification. From this pilot study, we found that MoveNet is the best human body pose detection while AlexNet is the best for classification.