Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Jan 22:15:1518294.
doi: 10.3389/fpls.2024.1518294. eCollection 2024.

YOLOv8s-Longan: a lightweight detection method for the longan fruit-picking UAV

Jun Li  1   2   3 , Kaixuan Wu  1 , Meiqi Zhang  1 , Hengxu Chen  1 , Hengyi Lin  1 , Yuju Mai  1 , Linlin Shi  1
Affiliations

YOLOv8s-Longan: a lightweight detection method for the longan fruit-picking UAV

Jun Li et al. Front Plant Sci. .

Abstract

Introduction: Due to the limited computing power and fast flight speed of the picking of unmanned aerial vehicles (UAVs), it is important to design a quick and accurate detecting algorithm to obtain the fruit position.

Methods: This paper proposes a lightweight deep learning algorithm, named YOLOv8s-Longan, to improve the detection accuracy and reduce the number of model parameters for fruitpicking UAVs. To make the network lightweight and improve its generalization performance, the Average and Max pooling attention (AMA) attention module is designed and integrated into the DenseAMA and C2f-Faster-AMA modules on the proposed backbone network. To improve the detection accuracy, a crossstage local network structure VOVGSCSPC module is designed, which can help the model better understand the information of the image through multiscale feature fusion and improve the perception and expression ability of the model. Meanwhile, the novel Inner-SIoU loss function is proposed as the loss function of the target bounding box.

Results and discussion: The experimental results show that the proposed algorithm has good detection ability for densely distributed and mutually occluded longan string fruit under complex backgrounds with a mAP@0.5 of 84.3%. Compared with other YOLOv8 models, the improved model of mAP@0.5 improves by 3.9% and reduces the number of parameters by 20.3%. It satisfies the high accuracy and fast detection requirements for fruit detection in fruit-picking UAV scenarios.

Keywords: YOLOv8-Longan network; attention mechanism; lightweight network; longan; target detection.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
The structure of the fruit-picking UAV. UAV, unmanned aerial vehicle.
Figure 2
Figure 2
YOLOv8s-Longan network structure.
Figure 3
Figure 3
Structure of the AMA attention module. AMA, Average and Max pooling attention.
Figure 4
Figure 4
Schematic diagram of the DenseNet structure.
Figure 5
Figure 5
Local structure diagram of the DenseAMA module. (A) DenseLayerAMA. (B) TransitAMA.
Figure 6
Figure 6
The specifics of the C2f-Faster-AMA module. (A) Faster-Block module structure. (B) Conventional Cony. (C) PCony. (D) C2f-Faster-AMA module structure.
Figure 7
Figure 7
VOVGSCSP module structure. (A) GSBottleneck. (B) VOVGSCSPC.
Figure 8
Figure 8
The specifics of the Inner-SIoU. (A) Schematic diagram. (B) Angle loss.
Figure 9
Figure 9
Data size of each category and label size distribution. (A) Size distribution of labels. (B) Heatmap of label size distribution. (C) Label size distribution graph.
Figure 10
Figure 10
Comparison of object detection algorithms’ indicators. (A) Precision comparison. (B) Recall comparison. (C) mAP@0.5 comparison. (D) mAP@0.5-0.95 comparison.
Figure 11
Figure 11
Prediction comparison of different network models for identification. (A) YOLOv5n. (B) YOLOv6n. (C) YOLOv8n. (D) YOLOv8n. (E) YOLOv6s. (F) YOLOv8s. (G) YOLOv8s-Longan detention results in dense longan string fruit scene.
Figure 12
Figure 12
Comparison of heatmaps for prediction of longan fruit recognition by different network models. (A) YOLOv5n. (B) YOLOv6n. (C) YOLOv8n. (D) YOLOv8n. (E) YOLOv6s. (F) YOLOv8s. (G) YOLOv8s-Longan heat map for longan string fruit prediction.
Figure 13
Figure 13
Comparison of detection results under different environmental conditions. (A) Far and sunny side. (B) Near and sunny side. (C) Far and night side. (D) Near and night side.
Figure 14
Figure 14
Comparison of the detection results for different longan varieties. (A) Far and Chuliang Longan. (B) Near and Chuliang Longan. (C) Far and Shixia Longan. (D) Near and Shixia Longan.
Figure 15
Figure 15
The UAV test scenario. UAV, unmanned aerial vehicle.
See this image and copyright information in PMC

References

    1. Cai L., Li H., Dong W., Fang H. (2022). Micro-expression recognition using 3d densenet fused squeeze-and-excitation networks. Appl. Soft Comput. 119. doi: 10.1016/j.asoc.2022.108594 - DOI
    1. Chen H., Chen H., Huang X., Zhang S., Chen S., Cen F., et al. . (2024. a). Estimation of sorghum seedling number from drone image based on support vector machine and yolo algorithms. Front. Plant Sci. 15, 1399872. doi: 10.3389/fpls.2024.1399872 - DOI - PMC - PubMed
    1. Chen W., Liu M., Zhao C., Li X., Wang Y. (2024. b). Mtd-yolo: Multi-task deep convolutional neural network for cherry tomato fruit bunch maturity detection. Comput. Electron. Agric. 216, 108533. doi: 10.1016/j.compag.2023.108533 - DOI
    1. Dairath M. H., Akram M. W., Mehmood M. A., Sarwar H. U., Akram M. Z., Omar M. M., et al. . (2023). Computer vision-based prototype robotic picking cum grading system for fruits. Smart Agric. Technol. 4, 100210. doi: 10.1016/j.atech.2023.100210 - DOI
    1. Ding Y., Zhang Z., Hu H., He F., Cheng S., Zhang Y. (2024). “Multi-feature fusion: Graph neural network and cnn combining for hyperspectral image classification,” In: Graph Neural Network for Feature Extraction and Classification of Hyperspectral Remote Sensing Images. (Springer, Singapore: ).

LinkOut - more resources