{"title":"Research on improvement strategies for a lightweight multi-object weed detection network based on YOLOv5","authors":"","doi":"10.1016/j.cropro.2024.106912","DOIUrl":null,"url":null,"abstract":"<div><p>Traditional weed detection technology has several limitations, including low detection accuracy, substantial computational demands, and large-scale detection models. To meet the requirements of weed multi-target identification and portability, this study proposes the YOLO–WEED model for weed recognition. The proposed model has the following innovations: (1) The backbone standard convolution module in YOLOv5 was replaced by the lightweight MobileNetv3 network to simplify the network structure and reduce parameter complexity; (2) The addition of convolutional block attention module (CBAM) to the neck network enabled the model to focus on the most important features while filtering out noise and irrelevant information; (3) To further improve classification accuracy and reduce loss, the C2f module was employed to improve the C3 module in the neck network; and (4) During the model plot process, a coordinate variable was added in the box label to help the model accurately locate the weeds. In the study, six species of weeds and one crop were used as test subjects. After image enhancement techniques were used, ablation experiments were deployed. The experimental results indicated that the YOLO–WEED model achieved an average accuracy of 92.5% in identifying six types of weeds and one type of crop. The accuracies for each type of plant were 82.7%, 97.3%, 98.8%, 86%, 93.5%, 99.3% and 89.6%, respectively. The number of model parameters was reduced by 39.4% compared with YOLOv5s. Furthermore, the localisation, classification and object losses were reduced by 0.025, 0.005 and 0.014, respectively. The model optimisation and deployment of the Jetson mobile terminal for multi-target detection were realised, and the performance was better than six network models such as YOLOv5s.</p></div>","PeriodicalId":10785,"journal":{"name":"Crop Protection","volume":null,"pages":null},"PeriodicalIF":2.5000,"publicationDate":"2024-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Crop Protection","FirstCategoryId":"97","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0261219424003405","RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"AGRONOMY","Score":null,"Total":0}
引用次数: 0
Abstract
Traditional weed detection technology has several limitations, including low detection accuracy, substantial computational demands, and large-scale detection models. To meet the requirements of weed multi-target identification and portability, this study proposes the YOLO–WEED model for weed recognition. The proposed model has the following innovations: (1) The backbone standard convolution module in YOLOv5 was replaced by the lightweight MobileNetv3 network to simplify the network structure and reduce parameter complexity; (2) The addition of convolutional block attention module (CBAM) to the neck network enabled the model to focus on the most important features while filtering out noise and irrelevant information; (3) To further improve classification accuracy and reduce loss, the C2f module was employed to improve the C3 module in the neck network; and (4) During the model plot process, a coordinate variable was added in the box label to help the model accurately locate the weeds. In the study, six species of weeds and one crop were used as test subjects. After image enhancement techniques were used, ablation experiments were deployed. The experimental results indicated that the YOLO–WEED model achieved an average accuracy of 92.5% in identifying six types of weeds and one type of crop. The accuracies for each type of plant were 82.7%, 97.3%, 98.8%, 86%, 93.5%, 99.3% and 89.6%, respectively. The number of model parameters was reduced by 39.4% compared with YOLOv5s. Furthermore, the localisation, classification and object losses were reduced by 0.025, 0.005 and 0.014, respectively. The model optimisation and deployment of the Jetson mobile terminal for multi-target detection were realised, and the performance was better than six network models such as YOLOv5s.
期刊介绍:
The Editors of Crop Protection especially welcome papers describing an interdisciplinary approach showing how different control strategies can be integrated into practical pest management programs, covering high and low input agricultural systems worldwide. Crop Protection particularly emphasizes the practical aspects of control in the field and for protected crops, and includes work which may lead in the near future to more effective control. The journal does not duplicate the many existing excellent biological science journals, which deal mainly with the more fundamental aspects of plant pathology, applied zoology and weed science. Crop Protection covers all practical aspects of pest, disease and weed control, including the following topics:
-Abiotic damage-
Agronomic control methods-
Assessment of pest and disease damage-
Molecular methods for the detection and assessment of pests and diseases-
Biological control-
Biorational pesticides-
Control of animal pests of world crops-
Control of diseases of crop plants caused by microorganisms-
Control of weeds and integrated management-
Economic considerations-
Effects of plant growth regulators-
Environmental benefits of reduced pesticide use-
Environmental effects of pesticides-
Epidemiology of pests and diseases in relation to control-
GM Crops, and genetic engineering applications-
Importance and control of postharvest crop losses-
Integrated control-
Interrelationships and compatibility among different control strategies-
Invasive species as they relate to implications for crop protection-
Pesticide application methods-
Pest management-
Phytobiomes for pest and disease control-
Resistance management-
Sampling and monitoring schemes for diseases, nematodes, pests and weeds.