DEVELOPMENT OF OPEN-SOURCE DETECTION ALGORITHM LANDMARKS FOR ANTHROPOMETRIC MEASUREMENTS AUTOMATIC ON HUMAN BODY MODEL RESULT THREE-DIMENSIONAL SCANNING

• Main Author: Hanan Atikah, Halima
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/85698
• Institution: Institut Teknologi Bandung

Traditional anthropometric measurement is a contact-based method that is time- consuming during data collection. With the advancement of three-dimensional scanning tools, non-contact anthropometric measurement methods have evolved into digital anthropometric measurements using 3D scans of the human body. As these measurement methods continue to develop, they are expected to enhance the accuracy and precision of anthropometric measurements efficiently. One approach to achieving this is through the automatic detection of landmarks using open-source software. Therefore, this study developed an algorithm for anthropometric measurement. Two methods were compared: the digital manual method using Blender and MATLAB software, and an open-source algorithm method using Python software. Anthropometric measurements were conducted on six subjects (3 males and 3 females) across 16 different body dimensions. Each subject was measured once using the algorithmic method and three times with the digital manual method. The algorithm's performance was evaluated by comparing it to the digital manual anthropometric measurements, which were considered the standard values. The accuracy of both methods was assessed using relative error. Most dimensions in the comparison between the two methods had relative errors of less than 5%, except for dimensions D, H, L, and M. Reliability was then measured using the Intraclass Correlation Coefficient (ICC). Of the 16 dimensions measured, 12 dimensions had ICC (A,1) values greater than 0.75, indicating good reliability. The upper and lower bounds of the 95% confidence interval based on the ICC values were also calculated. Based on the width of the 95% confidence interval, dimensions B, C, F, I, and N exhibited good reliability in terms of ICC values and stability estimation due to smaller uncertainty. Keywords: anthropometry, 3D model, algorithm, open-source, evaluation.