移动设备上YOLO模型的改进

YOLO is one of the most advanced real-time object detection networks. It can perform real-time object detection tasks on computing resource-rich devices, and ensure accuracy and speed. Because the computing resources of mobile devices are limited and YOLO needs more computing resources, YOLO can not achieve accurate and real-time detection on mobile devices. In this paper, we mainly study the optimization of YOLO neural network to keep its accuracy on the premise of relying on a small amount of computing resources. Finally, based on the improved YOLO, this paper builds an application program on iOS, and verifies the accuracy and fast performance of the improved YOLO on mobile devices.