<![CDATA[The global fashion e-commerce industry is currently facing an operational crisis due to high product return rates triggered by the mismatch between physical clothing sizes and consumer expectations. Conventional labeling systems (S, M, L, XL) have proven inadequate in representing the diversity of human anthropometry universally. This study proposes an innovative solution through the implementation of AI Pose Estimation using the MediaPipe framework and OpenCV for contactless body dimension extraction. The research method involves mapping 33 body landmarks, pixel-to-metric calibration based on user height, and body circumference estimation using the integration of Canny edge detection and Ramanujan II ellipse formulation. Scientific findings indicate that the system can achieve an average accuracy of 98.85% with a Mean Absolute Error (MAE) of only 0.88 cm across various body profiles. The conclusion of this study confirms that the use of Computer Vision-based Pose Estimation can improve the efficiency of the e-commerce industry by providing personalized and accurate size recommendations, which can systematically reduce product return rates and increase consumer loyalty through a more precise shopping experience.]]>
Copyrights © 2026
