Analyzing a Comprehensive Hand-Size Database Using Automated Landmark Detection and Measurement Techniques

Data derived from measurements of human hands are widely utilized across various fields, including virtual reality, medical science, fashion, and the development of hand size systems. The precision of hand size determination critically depends on the accurate extraction of measurement landmarks. This study emphasizes the importance of identifying primary dimensions that influence hand size systems and contributes to the development of a comprehensive hand-size database. We propose a method for measuring 24 distinct hand size parameters using 2D techniques. Our approach leverages the MediaPipe tool for accurate and automatic extraction of hand landmarks, regardless of skin color, alongside automatic hand border detection. The system was calibrated using Level 1-gauge blocks, achieving a relative error of 1.15%. Data were collected from over 500 workers, analyzing both right- and left-hand measurements for eight different male hand sizes, achieving a service rate of 70.3% in this study. This research provides a valuable tool for developing a worker hand size system, facilitating the manufacture of safety gloves and hand tools by offering an automated dataset for hand sizes.