基于改进YOLOv8的复杂果园环境下杏果的目标检测

doi:10.13733/j.jcam.issn.2095-5553.2025.03.036

摘要/Abstract

摘要：

为解决复杂果园环境中，因遮挡、重叠导致杏果识别误检率较高、检测精度较低的问题，提出一种基于改进YOLOv8n网络模型的杏果检测算法。该算法采用轻量化模块MobileViT—XS替换原有骨干网络，保证特征提取能力，同时降低模型的参数量与计算量，并且将原始的损失函数CIoU替换为WIoUv3，动态优化损失权重提高模型的检测精度。为验证改进方法的有效性，选取6种主流的目标检测模型、5种骨干网络的轻量化改进模型以及5种不同的损失函数进行对比试验。结果表明，改进后的模型相比原始模型在F1、平均精度均值mAP上提升1.25%、1.48%，参数量、浮点运算量、模型大小分别降低28.06%、0.1 G、1.48 MB。改进后的算法能够精准、快速地在复杂的果园环境中识别出杏果。

关键词: 杏果, 目标检测, 复杂果园环境, 轻量化网络, YOLOv8算法, 损失函数

Abstract:

This study addresses the challenges of high false detection rates and low accuracy in apricot recognition within complex orchard environments due to occlusion and overlap. An apricot detection algorithm based on an improved YOLOv8n network model is proposed. The algorithm replaces the original backbone network with the lightweight MobileViT—XS module, maintaining feature extraction capabilities while reducing the number of parameters and computational demands. The original loss function CIoU is substituted with WIoUv3, which dynamically optimizes loss weights and improves detection precision. Comparative experiments are conducted using six mainstream target detection models, five lightweight improved models with backbone networks, and five different loss functions. The results show that the improved model increases the F1 score and mAP by 1.25% and 1.48%, respectively, and reduces parameters, FLOPs, and model size by 28.06%, 0.1 G and 1.48 MB. This improved algorithm can accurately and quickly detect apricots in complex orchard environments.

Key words: apricots, target detection, complex orchard environment, lightweight network, YOLOv8 algorithm, loss function

中图分类号:

S662.2

买买提·沙吾提, 阿尔庆·西力克. 基于改进YOLOv8的复杂果园环境下杏果的目标检测[J]. 中国农机化学报, 2025, 46(3): 246-252.

Mamat Sawut, , , Aerqing Xilike . Target detection of apricots in complex orchard environments based on improved YOLOv8[J]. Journal of Chinese Agricultural Mechanization, 2025, 46(3): 246-252.

参考文献

［1］　吕春晶, 章秋平, 刘宁,等. 杏果核物理特性与其主要化学组分的相关性分析［J］.果树学报, 2021, 38（10）: 1717-1724.
Lü Chunjing, Zhang Qiuping, Liu Ning, et al. Correlations between physical properties and major chemical components of shells in apricot ［J］. Journal of Fruit Science, 2021,38（10）:1717-1724.
［2］　散鋆龙, 杨会民, 王学农,等. 振动方式和频率对杏树振动采收响应的影响［J］. 农业工程学报, 2018, 34（8）: 10-17.
San Yunlong, Yang Huimin, Wang Xuenong, et al. Effects of vibration mode and frequency on vibration harvesting of apricot trees ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2018, 34（8）: 10-17.
［3］　闫彬, 樊攀, 王美茸,等. 基于改进YOLOv5m的采摘机器人苹果采摘方式实时识别［J］. 农业机械学报, 2022,53（9）:28-38,59.
Yan Bin, Fan Pan, Wang Meirong,et al. Realtime apple picking pattern recognition for picking robot based on improved YOLOv5m ［J］. Transactions of the Chinese Society for Agricultural Machinery,2022,53（9）:28-38,59.
［4］　Lin G, Tang Y, Zou X, et al. Fruit detection in natural environment using partial shape matching and probabilistic Hough transform ［J］. Precision Agriculture, 2020, 21: 160-177.
［5］　Liu T H, Ehsani R, Toudeshki A, et al. Detection of citrus fruit and tree trunks in natural environments using a multielliptical boundary model ［J］. Computers in Industry, 2018, 99: 9-16.
［6］　廖崴, 郑立华, 李民赞,等. 基于随机森林算法的自然光照条件下绿色苹果识别［J］. 农业机械学报, 2017, 48（S1）:86-91.
Liao Wei, Zhen Lihua, Li Minzan, et al. Green apple recognition in natural illumination based on random forest algorithm ［J］. Transactions of the Chinese Society for Agricultural Machinery,2017,48（S1）:86-91.
［7］　Wan S, Goudos S. Faster R—CNN for multiclass fruit detection using a robotic vision system ［J］. Computer Networks, 2020, 168: 107036.
［8］　彭红星, 黄博, 邵园园,等. 自然环境下多类水果采摘目标识别的通用改进SSD模型［J］. 农业工程学报, 2018,34（16）:155-62.
Peng Hongxing, Huang Bo, Shao Yuanyuan,et al. General improved SSD model for picking object recognition of multiple fruits in natural environment ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2018, 34（16）: 155-162.
［9］　赵辉, 乔艳军, 王红君,等. 基于改进YOLOv3的果园复杂环境下苹果果实识别［J］. 农业工程学报, 2021, 37（16）:127-35.
Zhao Hui, Qiao Yanjun, Wang Hongjun,et al. Apple fruit recognition in complex orchard environment based on improved YOLOv3 ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37（16）: 127-135.
［10］　赵凯琳, 靳小龙, 王元卓. 小样本学习研究综述［J］. 软件学报,2021,32（2）:349-369.
Zhao Kailin, Jin Xiaolong, Wang Yuanzhuo. Survey on fewshot learning ［J］. Journal of Software, 2021, 32（2）:349-369.
［11］　Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection ［C］. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 2117-2125.
［12］　Liu S, Qi L, Qin H, et al. Path aggregation network for instance segmentation ［C］. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 8759-8768.
［13］　彭炫, 周建平, 许燕,等. 改进YOLOv5识别复杂环境下棉花顶芽［J］. 农业工程学报， 2023, 39（16）: 191-197.

Peng Xuan, Zhou Jianping, Xu Yan,et al. Cotton top bud recognition method based on YOLOv5-CPP in complex environment ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39（16）: 191-197.

［14］　彭书博, 陈兵旗, 李景彬,等. 基于改进YOLOv7的果园行间导航线检测［J］. 农业工程学报， 2023,39（16）131-138.
Peng Shubo, Chen Bingqi, Li Jingbin, et al. Detection of the navigation line between lines in orchard using improved YOLOv7［J］. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39（16）: 131-138.
［15］　Li X, Wang W, Wu L, et al. Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection ［J］. Advances in Neural Information Processing Systems, 2020, 33: 21002-12.
［16］　Mehta S, Rastegari M. MobileViT: Lightweight, generalpurpose, and mobilefriendly vision transformer ［J］. arXiv preprint arXiv:211002178, 2021.
［17］　Sandler M, Howard A, Zhu M, et al. MobileNetV2: Inverted residuals and linear bottlenecks ［C］. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 4510-4520.
［18］　Zheng Z, Wang P, Liu W, et al. DistanceIoU loss: Faster and better learning for bounding box regression ［C］. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34（7）: 12993-13000.
［19］　Zhang Y F, Ren W, Zhang Z, et al. Focal and efficient IoU loss for accurate bounding box regression ［J］. Neurocomputing, 2022, 506: 146-157.
［20］　Gevorgyan Z. SIoU loss: More powerful learning for bounding box regression ［J］. arXiv preprint arXiv:220512740, 2022.
［21］　Tong Z, Chen Y, Xu Z, et al. WiseIoU: Bounding box regression loss with dynamic focusing mechanism ［J］.arXiv preprint arXiv:230110051, 2023.

[1]	吕宗旺, 王甜甜, 孙福艳, 祝玉华, . 基于改进YOLOv8m的小麦仓储粮虫检测方法[J]. 中国农机化学报, 2025, 46(3): 108-114.
[2]	郭敬涛, 吕凤, 章慧婷, 杨彪, 刘大洋. 基于迁移学习与轻量化YOLOv5s的草莓目标检测方法[J]. 中国农机化学报, 2025, 46(3): 253-260.
[3]	许毓超, , 吴茜, , 张兵园, , , 周玲莉, , 任妮, , , 张美娜, , . 轻量级深度学习网络在农作物目标检测的应用进展[J]. 中国农机化学报, 2025, 46(3): 261-270.
[4]	罗柳茗, 李岩舟, 石美琦, 黄鑫, 陈汐 . 基于YOLOv8n的甘蔗杂草识别方法 [J]. 中国农机化学报, 2025, 46(2): 237-244.
[5]	夏子林, 张新洲, 王文波, 夏先飞, 陈兰, 顾寄南. 基于机器视觉的蚕豆荚高精度检测方法研究[J]. 中国农机化学报, 2025, 46(1): 157-163.
[6]	彭勇, 乔印虎, 张春燕, 姚杰, 包殿令. 改进YOLOv5模型的草莓检测方法研究[J]. 中国农机化学报, 2025, 46(1): 213-219.
[7]	张传栋, 亓璐, 丁华立. 基于改进 YOLOv8n模型的多品种葡萄簇检测方法[J]. 中国农机化学报, 2024, 45(9): 220-226.
[8]	李邦国, 王辉, 宋杨, 任志伟, 刘跃华, 徐乐程. 基于无人驾驶小麦收割机立体视觉感知系统[J]. 中国农机化学报, 2024, 45(9): 244-249.
[9]	赵晓霞, 程曼, 袁洪波. 基于StrongSORT算法的羊只多目标跟踪方法[[J]. 中国农机化学报, 2024, 45(8): 180-188.
[10]	冯晓, 张辉, 刘运超, 张微, 李小红, 马中杰, . 基于改进YOLOv5s的硬核期葡萄簇检测[J]. 中国农机化学报, 2024, 45(8): 240-245.
[11]	周思捷, 刘天奇, 陈天华. 基于改进YOLOv5算法的水稻病害识别研究[J]. 中国农机化学报, 2024, 45(8): 246-253.
[12]	岳耀华, 张伟, 亓立强, . 基于改进YOLOv5的田间大豆花朵生长状态检测方法研究[J]. 中国农机化学报, 2024, 45(7): 188-193.
[13]	张延军, 赵建鑫, . 基于多路神经网络的黄花菜关节点检测算法研究[J]. 中国农机化学报, 2024, 45(7): 228-234.
[14]	张晨一, 张晓乾, 任振辉. 基于改进YOLOv3的自然场景中未成熟青苹果识别研究[J]. 中国农机化学报, 2024, 45(7): 243-248.
[15]	薛霞, 刘鹏, 周文. 基于改进YOLOv8算法的实时细粒度植物病害检测[J]. 中国农机化学报, 2024, 45(5): 188-194.