基于深度学习的轻量化草莓成熟度检测模型

doi:10.14088/j.cnki.issn0439-8114.2026.04.002

摘要/Abstract

摘要： 为实现复杂环境下草莓成熟度的实时检测,提出一种基于YOLOv5s改进的轻量化目标检测模型YOLOv5s-DCC,以期达到精度和轻量化的双重提升。将深度可分离卷积嵌入骨干网络(Backbone)结构中降低计算复杂度,在颈部网络(Neck)部分引入CARAFE上采样增强对细小特征捕获的能力,在检测头(Head)部分融合CBAM注意力机制,通过动态权重提升特征选择性能,构建了兼顾检测精度与轻量化的优化模型。改进后的模型大小仅为13.9 MB,精确率、召回率、平均精度(mAP)分别为93.4%、92.7%、95.3%。相比于原模型YOLOv5s,精确率、召回率、平均精度分别提高1.3、0.9和1.6个百分点,参数量减少0.5×10⁶。相较于YOLOX-s、YOLOv7-Tiny和YOLOv8s等主流轻量化模型,YOLOv5s-DCC综合性能最优,能够满足复杂环境下农业采摘机器人对草莓果实成熟度的实时检测需求。

关键词: 草莓成熟度检测, 轻量化模型, 深度可分离卷积, CBAM注意力机制, CARAFE上采样

Abstract: To achieve real-time detection of strawberry ripeness in complex environments, a lightweight object detection model YOLOv5s-DCC based on an enhanced YOLOv5s architecture was proposed, aiming to enhance both accuracy and computational efficiency. Depthwise separable convolutions were embedded within the backbone structure to reduce computational complexity. The neck section incorporated CARAFE upsampling to enhance the capture of minute features, while the detection head integrated CBAM attention mechanisms. Dynamic weighting improved feature selection performance, resulting in an optimized model balancing detection accuracy and lightweight efficiency. The refined model size was only 13.9 MB, achieving precision, recall, and mean average precision (mAP) of 93.4%, 92.7%, and 95.3%, respectively. Compared to the original YOLOv5s model, precision, recall, and mAP improved by 1.3, 0.9, and 1.6 percentage points, respectively, while reducing parameters by 0.5×10⁶. Compared to mainstream lightweight models such as YOLOX-s, YOLOv7-tiny, and YOLOv8s, YOLOv5s-DCC deliverd the best overall performance. It could meet the real-time strawberry ripeness detection requirements of agricultural harvesting robots in complex environments.

Key words: strawberry ripeness detection, lightweight model, depth separable convolution, CBAM attention mechanism, CARAFE upsampling

中图分类号:

TP391.41

李慧琴, 王洋洋, 颉世国, 王鹏飞, 兰明明. 基于深度学习的轻量化草莓成熟度检测模型[J]. 湖北农业科学, 2026, 65(4): 7-15.

LI Hui-qin, WANG Yang-yang, XIE Shi-guo, WANG Peng-fei, LAN Ming-ming. A lightweight model for strawberry ripeness detection based on deep learning[J]. HUBEI AGRICULTURAL SCIENCES, 2026, 65(4): 7-15.

参考文献

[1] FAN Y C, ZHANG S Y, FENG K, et al.Strawberry maturity recognition algorithm combining dark channel enhancement and YOLOv5[J]. Sensors, 2022, 22(2): 419.
[2] 李东辉,余宏杰.基于改进YOLOv8n模型的草莓识别方法[J].安徽科技学院学报,2025,39(1):48-59.
[3] 梁敖,代东南,牛思琪,等.基于改进YOLOv5s的草莓成熟度实时检测算法[J].山东农业科学,2024,56(11):156-163.
[4] 何斌,张亦博,龚健林,等.基于改进YOLO v5的夜间温室番茄果实快速识别[J].农业机械学报,2022,53(5):201-208.
[5] 张晖耀,黄力湘,陈继清,等.基于改进YOLOv5s的草莓成熟度检测算法[J].南京农业大学学报,2025,48(4):990-999.
[6] 刘诗怡,胡滨,赵春.基于改进YOLOv7的黄瓜叶片病虫害检测与识别[J].农业工程学报,2023,39(15):163-171.
[7] BAI Y F, YU J Z, YANG S Q, et al.An improved YOLO algorithm for detecting flowers and fruits on strawberry seedlings[J]. Biosystems engineering, 2024, 237: 1-12.
[8] LUO Q, WU C B, WU G J, et al.A small target strawberry recognition method based on improved YOLOv8n model[J]. IEEE access, 2024, 12: 14987-14995.
[9] LI J J, ZHU Z F, LIU H X, et al.Strawberry R-CNN: Recognition and counting model of strawberry based on improved faster R-CNN[J]. Ecological informatics, 2023, 77: 102210.
[10] WANG J X,SU Y H, YAO J H, et al.Apple rapid recognition and processing method based on an improved version of YOLOv5[J]. Ecological informatics, 2023, 77: 102196.
[11] NAN Y L, ZHANG H C, ZENG Y, et al.Intelligent detection of Multi-Class pitaya fruits in target picking row based on WGB-YOLO network[J]. Computers and electronics in agriculture, 2023, 208: 107780.
[12] 姜业帆,江朝晖,朱瑞,等.基于PCIA-YOLO的复杂环境下草莓成熟度轻量化检测方法[J].南京农业大学学报,2025,48(6):1451-1463.
[13] 章璞,乔波,陈义明.基于改进YOLOv8的多阶段草莓检测算法[J].中国农机化学报,2024,45(10):274-280.
[14] WANG D, SHANG Y.A new active labeling method for deep learning[A]. 2014 International Joint Conference on Neural Networks (IJCNN)[C]. Beijing, China: IEEE, 2014.112-119.
[15] KHANAM R, HUSSAIN M. What is YOLOv5: A deep look into the internal features of the popular object detector[EB/OL].2024: arXiv: 2407.20892. https://arxiv.org/abs/2407.20892
[16] CHOLLET F.Xception: Deep learning with depthwise separable convolutions[A]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)[C]. Honolulu, HI, USA: IEEE, 2017.1800-1807.
[17] WANG J Q, CHEN K, XU R, et al.CARAFE: Content-aware ReAssembly of FEatures[A]. 2019 IEEE/CVF International Conference on Computer Vision (ICCV)[C]. Seoul, Korea: IEEE, 2020.3007-3016.
[18] 皇甫晓瑛,钱惠敏,黄敏.结合注意力机制的深度神经网络综述[J].计算机与现代化,2023(2):40-49,57.
[19] 苗长云,孙丹丹.基于改进YOLOv5s的带式输送机滚筒故障检测研究[J].工矿自动化,2023,49(7):41-48.
[20] 王法安,何忠平,张兆国,等.基于PN-YOLO v8s-Pruned的轻量化三七收获目标检测方法[J].农业机械学报,2024,55(11):171-183.
[21] JIN X, XIE Y P, WEI X S, et al.Delving deep into spatial pooling for squeeze-and-excitation networks[J]. Pattern recognition, 2022, 121: 108159.
[22] WANG Q L, WU B G, ZHU P F, et al.ECA-net: Efficient channel attention for deep convolutional neural networks[A]. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)[C]. Seattle,WA,USA: IEEE, 2020.11531-11539.
[23] HOU Q B, ZHOU D Q, FENG J S.Coordinate attention for efficient mobile network design[A]. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)[C]. Nashville, TN, USA: IEEE, 2021.13708-13717.
[24] WOO S, PARK J, LEE J Y, et al.CBAM: Convolutional block attention module[A]. FERRARI V, HEBERT M, SMINCHISESCU C, et al, eds. Computer Vision-ECCV 2018[C]. Cham: Springer, 2018.3-19.