This is a computer vision model based on yolov3 in python. The model was trained by manually drawing bounding boxes around the hatch for 1000+ pictures and left to train overnight using Supervisely.