Development of Contactless Anthropometric Body Measurement Using MediaPipe

Anthropometric body measurement involves measuring the dimensions and proportions of the human body, and has important applications in various fields, including fashion, sports, and medicine. The clothing industry's value-chain of clothing and garments for ready-to-wear clothes to customized apparels mostly rely on manual measurement to determine client's body's anthropometrics. This time-consuming method is prone to human error, which can result in the need for garment readjustments. This study presents the development of a contactless body measurement system using a computer vision approach and MediaPipe Pose Landmarker to locate key points for pose body estimates. The system performs human body detection and captures images from four different angles and then extract measurements of specific body parts, including shoulder length, arm length, upper arm length, hip girth, hip height, thigh girth, knee girth, knee height, calf girth, and ankle girth. With the web application developed, users can upload self-taken photos to the system which are then processed by a body measurement algorithm. Experimentation done with 10 subjects; the system achieved an overall accuracy of 88.67% when compared against manual data. The linear measurements yield higher accuracy of 92.16% as compared to girth measurements of 85.17% due to the complexity of the 3D shape of the human body. The web application may provide a seamless platform for online processing of apparel requests and transactions. Overall, the system may significantly improve precision and efficiency in the tailoring process for enterprises in the apparel industry.