Object recognition in mobile applications with artificial intelligence technologies

With the development of technology, it is seen that almost everyone has a smart mobile device. In this thesis study, it is aimed to recognize and detect the objects in the image by using various techniques, regardless of whether the hardware capabilities of smart mobile devices are high or not. YOLO, one of the current object detection algorithms based on deep learning, is an object detection tool that has been under development since it was first introduced. YOLO has become the preferred tool for object detection problems in the commercial field due to its speed and accuracy. In this thesis, an interface for the smart mobile device and a server have been developed for the YOLOv7 model to realize the object detection problem. The 7th version of YOLO, YOLOv7, was preferred in this study because it achieved an average accuracy of 51.2% after training with the Microsoft COCO dataset. has been done. While high-capacity hardware is generally needed to run deep learning-based systems, the object detection problem has been successfully solved with this designed server, regardless of the graphics processing unit (GPU) of a smart mobile device.