Estimating the height of a person from a video sequence

Abstract

The research described in this paper focuses on a development of experimental solution, which estimates human height from a video sequence. The main requirements include portability, low hardware requirements, and accuracy. The paper includes analysis of the algorithms employed for image processing, human detection and methods for estimating height of human. The experimental software solution is implemented in Python programming language. The SSD algorithm was selected to detect the person in the image and the output was later on utilized to calculate the real human height. A calibration object in a form of ArUco marker and its known dimensions became the foundation for calculating the real human height. The experimental testing confirmed the accuracy of the developed solution to be sufficient for the real-world application. The most accurate height, with just a deviation of 0.001 meters, was estimated at a distance of 5 and 8 meters. The approach presented in this paper is also to be used as a complement by newly proposed unmanned aerial vehicle security system.