Measurement of Human Height by using the Convolutional Neural Network and the Trigonometric Theory
Rungnirun Jindarat, Thanomsak Sopon, Kidsanapong Puntsri, Sasiphan Wongtrairat, Wannaree Wongtrairat · 2022
Measurement of human height from a digital camera using the technique of object detection to select a sub-image of interest by pointing the target on both feet and the human as keypoint detection by using U-Net convolutional neural network. These keypoint was used in conjunction with the relationship of triangles. It will help to be able to calculate the height of human accurately. From the experimental results when tested with three humans 162,173 and 180 cm, the maximum error is 2.76%. the percent of error increase when the distance between the human and the camera increases. Moreover, the height of the camera affects the error value. The percent of error will high when the height human.