Uses a convolutional Neural Network with residual aspects and a You Only Look Once loss format to generate the image coordinates of a box's corners. Fundamentals of epipolar and regular geometry are utilized to analyze two images from a stereoscopic camera, calculating the x, y and z coordinates in the real world of each of these corners. From there, the program draws vectors between each of the edges of this box and calculates the magnitude, providing the edge length in the real world.