AUTHOR=Li He , Guo Changle , Yang Zishang , Chai Jiajun , Shi Yunhui , Liu Jiawei , Zhang Kaifei , Liu Daoqi , Xu Yufei 

TITLE=Design of field real-time target spraying system based on improved YOLOv5

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.1072631

DOI=10.3389/fpls.2022.1072631

ISSN=1664-462X

ABSTRACT=Target spraying is a highly effective application technology that can effectively improve pesticide utilization and reduce pesticide hazards. It uses image processing and sensor technology to achieve crop target identification and precise spraying by automatically controlling the opening and closing of the nozzle. In the past, it was often applied to scenarios with large targets such as fruit trees, while for small targets, dense targets and field environments with complex backgrounds, crop recognition algorithms based on traditional image processing algorithms performed poorly in terms of accuracy and robustness. This project combines deep learning algorithms with precision spraying technology to design a machine vision precision real-time targeting spraying system for field scenarios. Firstly, the overall system design was proposed for the real-time target spraying system, including the image acquisition and detection module, the electronically controlled spraying module, and the stable pressure drug supply module. After that, based on the target detection model YOLOv5s, the model is lightened and improved by replacing the backbone network and adding an attention mechanism. Based on this, a grille decision control algorithm for solenoid valve group on-off was designed, while common malignant weeds were selected as objects to produce data sets and complete model training. Finally, the deployment of the hardware system and detection model on the electric spray bar sprayer was completed, and field trials were conducted at different speeds. The experimental results show that the improved algorithm reduces the model size to 53.57% of the original model with less impact on mAP accuracy, improves FPS by 18.16%, and can detect weeds accurately under complex situations such as overexposure, shading, and shadow interference. The accuracy of on-target spraying at 2km/h, 3km/h and 4km/h speeds were 90.80%, 86.20% and 79.61%, respectively.