Leveraging OpenPose and Kinect: Cutting-edge technologies for Ergonomic Risk Assessment
Year: 2025
Type: article
Abstract: Motion analysis is essential in industrial settings for optimizing production, improving efficiency, and ensuring worker safety. Technologies such as Azure Kinect and OpenPose facilitate movement monitoring through image and video processing. This study compares two pose estimation methods by analysing joint angles – including the knee, elbow, neck, hip, and shoulder – during tasks performed by six participants. Additionally, three applications involving different activities were evaluated to assess system reliability in workplace contexts, with three participants performing the tasks. Results indicate that both systems detect angles accurately, with acceptable variation ranges across most tasks analysed. However, OpenPose showed low accuracy in capturing angles in the frontal view, highlighting a limitation of its 2D joint detection with frontal perspective and occlusions. For tasks involving frontal views or manual handling of large objects, we recommended cross-verifying OpenPose’s angle data, as Azure Kinect provides more consistent accuracy in these conditions.
Source: IFAC-PapersOnLine
Institution: University of Modena and Reggio Emilia
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4414568331