Vision-Based Computing Pipeline for Recognizing Hand Grip-Types During Tool Handling

OCCUPATIONAL APPLICATIONSHands are frequently exposed to the risks of musculoskeletal disorders (MSDs) due to their involvement in tool handling. We propose a computer vision-based pipeline that can accurately recognize the types of hand grips during tool handling using monocular red, green, and blue (RGB) images. While the predominant methods for hand ergonomics assessments require considerable training for practitioners, the proposed pipeline can facilitate assessments by providing crucial information, such as grip types, duration, and repetition, in a continuous and noninvasive manner. The proposed pipeline could support preventative measures, early diagnosis, and effective treatments for hand-related MSDs in workspaces. Additionally, the simple setup of the proposed pipeline can be integrated with other ergonomics assessment tools that are not limited to the hands, contributing to a more comprehensive analysis of MSD risks across body parts.