Image Preprocessing for Efficient Training of YOLO Deep Learning Networks

Every artificial intelligence system needs big data for training. In particular, artificial intelligence for object detection requires a lot of images for training. It is possible to obtain training images from internet by web crawler. In most cases, however, images collected by the crawler are often not refined. In other words, it can't be used as training data in any artificial intelligence platform without proper pre-processing. This paper describes a pre-processing system for images collected by a crawler for YOLO training. And it will proved that this system is capable of efficient training.