基于改进YOLOv8模型的黄花菜花蕾识别研究

doi:10.14088/j.cnki.issn0439-8114.2025.07.032

摘要/Abstract

摘要： 通过深度融合CSPNet与密集连接网络（DenseNet）构建CSPDenseNet骨干模块,将该模块集成到YOLOv8模型,替换主干网络末端的最后2个标准卷积模块,得到改进YOLOv8模型（Dense-YOLOv8）。结果表明,在简单背景、稀疏黄花菜（Hemerocallis citrina Baroni）花蕾场景下,Dense-YOLOv8模型成功识别出全部成熟花蕾;在简单背景、密集黄花菜花蕾场景下,Dense-YOLOv8模型在花蕾检测任务中展现出优异的识别性能,但在处理紧密相邻目标时仍存在部分漏检现象;在复杂背景、密集黄花菜花蕾场景下,Dense-YOLOv8模型成功识别出全部成熟花蕾。Dense-YOLOv8模型的mAP、F1、识别速度、模型大小分别为90.75%、89%、53 f/s、217.68 MB;与YOLOv8模型、Faster R-CNN模型、YOLOv7相比,Dense-YOLOv8 模型在精简网络结构与参数的同时,显著提升了目标检测的精度与速度。

关键词: 改进YOLOv8模型, 深度学习, 黄花菜（Hemerocallis citrina Baroni）, 花蕾, 识别

Abstract: A CSPDenseNet backbone module was constructed by deeply integrating CSPNet and DenseNet. This module was integrated into the YOLOv8 model, replacing the last two standard convolutional modules at the end of the backbone network, resulting in the improved YOLOv8 model (Dense-YOLOv8).The results demonstrated that the Dense-YOLOv8 model successfully identified all mature buds under scenarios with a simple background and sparse daylily (Hemerocallis citrina) Baroni buds. Under scenarios with a simple background and dense daylily buds, the Dense-YOLOv8 model exhibited excellent recognition performance in the bud detection task, although some missed detections still occurred when processing tightly adjacent targets.Under scenarios with a complex background and dense daylily buds, the Dense-YOLOv8 model successfully identified all mature buds.The mAP, F1, recognition speed, and model size of the Dense-YOLOv8 model were 90.75%, 89%, 53 f/s, and 217.68 MB, respectively. Compared with the YOLOv8 model, Faster R-CNN model and YOLOv7 model, the Dense-YOLOv8 model significantly improved both the accuracy and speed of object detection while streamlining the network structure and reducing parameters.

Key words: improved YOLOv8 model, deep learning, daylily (Hemerocallis citrina Baroni), bud, recognition

中图分类号:

TP391

霍静琦, 崔婷婷, 薛志璐. 基于改进YOLOv8模型的黄花菜花蕾识别研究[J]. 湖北农业科学, 2025, 64(7): 186-191.

HUO Jing-qi, CUI Ting-ting, XUE Zhi-lu. Research on daylily buds recognition based on an improved YOLOv8 model[J]. HUBEI AGRICULTURAL SCIENCES, 2025, 64(7): 186-191.

参考文献

[1] LIU X, ZHAO D, JIA W, et al.A detection method for apple fruits based on color and shape features[J]. IEEE access, 2019, 7: 67923-67933.
[2] TAN W, ZHAO C, WU H.Intelligent alerting for fruit-melon lesion image based on momentum deep learning[J]. Multimedia tools and applications, 2016, 75(24): 16741-16761.
[3] TIAN Y, YANG G, WANG Z, et al.Apple detection during different growth stages in orchards using the improved YOLOv3 model[J]. Computers and electronics in agriculture, 2019, 157: 417-426.
[4] STEUN M, BARGOTI S, UNDERWOOD J.Image based mango fruit detection, localisation and yield estimation using multiple view geometry[J]. Sensors, 2016, 16(11): 1915.
[5] YU Y, ZHANG K, YANG L, et al.Fruit detection for strawberry harvesting robot in non-structural environment based on Mask-RCNN[J]. Computers and electronics in agriculture, 2019, 163: 104846.
[6] FU L, FENG Y, ELKAMIL T, et al.Image recognition method of multi-cluster kiwifruit in field based on convolutional neural networks[J]. Transactions of the Chinese society of agricultural engineering, 2018, 34(2): 205-211.
[7] 杨千. 平菇采摘机器人目标识别与定位控制关键技术研究[D].江苏苏州:苏州大学,2020.
[8] REN Z, LAM E Y, ZHAO J.Real-Time target detection in visual sensing environments using deep transfer learning and improved anchor box generation[J]. IEEE access, 2020, 8: 193512-193522.
[9] 熊俊涛,郑镇辉,梁嘉恩,等.基于改进YOLOv3网络的夜间环境柑橘识别方法[J].农业机械学报,2020,51(4):199-206.
[10] LIU Z, WU J, FU L, et al.Improved kiwifruit detection using pre-trained VGG16 with RGB and NIR information fusion[J]. IEEE access, 2019, 8: 2327-2336.
[11] LIU Y P, YANG C H, LING H, et al.A visual system of citrus picking robot using convolutional neural networks[A].2018 5th international conference on systems and informatics (ICSAI)[C]. Nanjing,China : IEEE, 2018. 344-349.
[12] REN S, HE K, GIRSHICK R, et al.Faster R-CNN: Towards real-time object detection with region proposal networks[J]. Advances in neural information processing systems, 2015, 28: 91-99.
[13] REDMON J, DIVVALA S, GIRSHICK R, et al.You only look once: Unified, real-time object detection[A].Proceedings of the IEEE conference on computer vision and pattern recognition[C].Las Vegas, USA : IEEE, 2016.779-788.
[14] WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[A].Proceedings of the IEEE/CVF conference on computer vision and pattern recognition[C].Vancouver,Canada:IEEE,2023.7464-7475.
[15] HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[A]. Proceedings of the IEEE conference on computer vision and pattern recognition[C].Hawaiian, USA: IEEE, 2017.4700-4708.
[16] WANG C Y, LIAO H Y M, Wu Y H, et al. CSPNet: A new backbone that can enhance learning capability of CNN[A].Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops[C]. Seattle,USA: IEEE,2020.390-391.