Study on YOLOv5 Algorithm for Fruit Sorting and Discriminating in Complex Scenes

It is of great significance to realize the classification and discriminant detection of fruit in the assembly line under complex environments, change the current situation of high cost, long time and uneven sorting of fruit manually, and improve the efficiency and accuracy of fruit detection. This study designed a discriminating system based on YOLOv5 that can automatically, efficiently and accurately sort fruits in complex scenarios with different light, partial overlap, multiple fruit still photos and video shooting situations. Raspberry PI 5 was equipped with YOLOv5 as the visual recognition part, and STM32f407 development board was used as the control system for pipeline sorting part. Receiving and processing the recognized data to control the stepper motor, digital steering gear, coding motor for sorting operation, not only has the function of recognizing the type and shape of fruit, but also can efficiently distinguish the superior and inferior fruits and classify them. With the position sensor, the sorting pick for different kinds of fruit sorting accuracy of more than 98.9%, because the system implementation method is relatively simple, greatly improve the sorting efficiency of fruit production line and fruit quality after sorting, reduce labor costs, promote the development of fruit industry, has a good market application prospects.