MECHANICAL OBJECT PARTS DETECTION USING DEEP LEARNING BASED YOLO MODELS

With the development of technology and artificial intelligence algorithms, we see that machines and smart systems take place in many areas. Especially in the industry, their use has started to become widespread day by day, as they make fewer mistakes than people and can produce more serially and with higher quality. The ability of machines to perform certain tasks by interacting with the outside world first depends on perceiving the objects in their environment. Detection processes of machines are realized with auxiliary tools such as sensors, switches and cameras. With deep learning, where more complex structures can be resolved compared to machine learning, studies in this direction continue to progress rapidly. In this study, it is aimed to determine the mechanical parts with the camera and to determine the number of the mechanical parts for the stock control and stock management of the companies engaged in mass production. One of the deep learning algorithms, Yolov5 is used for object detection and object counting. As a result of the study, the system works properly and successfully fulfills the specified functions