This study presents a non-contact method for estimating human height from a distance of 2 to 3 m during horizontal body rotations, eliminating the reliance on manual or traditional contact-based techniques. The proposed system leverages computer vision, integrating OpenCV and Mediapipe libraries with the YOLOv8 model. Mediapipe is used to detect and annotate key body joints, while YOLOv8 identifies these annotated points. By analyzing the ratio of distances between specific joints and applying a linear regression model, the system accurately estimates a person's height. Experimental results show that the method achieves a low error rate of approximately 1.2%. Future research will aim to generalize this approach to various body positions—such as standing, lying, sitting cross-legged, and bending the legs—as well as accommodate different distances and camera angles, thereby improving its versatility and accuracy.