On the Application of Human Pose Estimation in a Driver Condition Monitoring Task
Artem Pavlikov, Vsevolod Volkogonov, Anastasiia Lipatova
Abstract
Going on a trip, however, this does not guarantee the normal condition of the driver while driving. This paper proposes a method of monitoring the driver based on video images that are processed by a neural network and determines such indicators as head movement, arm movement, and upper shoulder joints. The paper compares the main architectures of neural network frameworks: AlphaPose, OpenPose, PoseNet and Mask R-CNN. Mathematical description of kinematics of human hands and head movement is given A program continuous monitoring of driver's condition by means of video image processing in real conditions is proposed.
Topics & Concepts
Task (project management)Computer sciencePoseEstimationArtificial intelligenceComputer visionEngineeringSystems engineeringGait Recognition and AnalysisHuman Pose and Action RecognitionOccupational Health and Performance