Fast Multi Object Detection and Counting by YOLO V3

Object detection (OD) has been a deep and vital field in different industries such as security, medical and automobile industry. Counting identified objects has been proved as a crucial research field. You Only Look Once (YOLO) algorithm is a very powerful algorithm for this aim with a quick performance compared to other existing algorithms. In this paper, an algorithm to detect and count objects simultaneously from a single photo is proposed based on YOLO V3 algorithm. The YOLO V3 algorithm has been modified to be able to classify 82 types of objects. After this algorithm succeeds in detecting objects, it draws bounding boxes around each object in the image and displays the exact type of them, and counts the bounding boxes drawn around the object, determines the number of each specific object. The photos are taken regularly from the environment and a kind of auto-shooting algorithm is employed. The results show that the algorithm presented in this paper succeeds in counting several objects in a single image with reduced calculation time and a very light process. It will be clearly visible that the approach of this article can help many industries that need to count different objects fast and cheaply.