[1]刘振宇,黄靖,陈梦飞,等.基于深度学习的双孢菇工业识别方法[J].福建理工大学学报,2025,23(06):591-598.[doi:10.3969/j.issn.2097-3853.2025.06.012]
　LIU Zhenyu,HUANG Jing,CHEN Mengfei,et al.Industrial identification method of Agaricus bisporus based on deep learning[J].Journal of Fujian University of Technology;,2025,23(06):591-598.[doi:10.3969/j.issn.2097-3853.2025.06.012]
点击复制

基于深度学习的双孢菇工业识别方法()
分享到：

《福建理工大学学报》[ISSN:2097-3853/CN:35-1351/Z]

卷:
第23卷
期数:
2025年06期
页码:
591-598
栏目:
出版日期:
2025-12-25

文章信息/Info

Title:
Industrial identification method of Agaricus bisporus based on deep learning
作者:
刘振宇; 黄靖; 陈梦飞; 翟振林; 连海俊; 叶荣坤
福建水利电力职业技术学院
Author(s):
LIU Zhenyu;  HUANG Jing;  CHEN Mengfei;  ZHAI Zhenlin;  LIAN Haijun;  YE Rongkun
Faculty of Electromechanical Engineering, Fujian College of Water Conservancy and Electric Power
关键词:
双孢菇识别; SIAF-YOLOv8; AFPN; SimAM三维注意力机制
Keywords:
identification of Agaricus bisporus; SIAF-YOLOv8; AFPN; SimAM 3D attention mechanism
分类号:
TP391.4；S2
DOI:
10.3969/j.issn.2097-3853.2025.06.012
文献标志码:
A
摘要:
针对工业化双孢菇培育中人工识别方法效率低、质量不稳定等问题，提出一种改进的SIAF-YOLOv8双孢菇识别算法，实现自动化精准检测需求。首先基于4 层渐近特征金字塔网络(AFPN)重构特征融合结构，通过多层级特征交互减少语义差距；其次在骨干网络嵌入SimAM三维注意力机制，实现无参量特征自适应加权。采用自建双孢菇数据集(2 111 幅图像，60 053 个样本)进行验证，结果表明:改进算法AP50 和AP50-95 分别达到98.5%和85.7%，较原YOLOv8n 提升1.0%、4.6%；模型权重减至5.6 MB，单图推理时间12 ms，对比当前几种通用目标检测算法，SIAF-YOLOv8 在检测精度、速度与模型轻量化方面实现最优平衡，可为农业机器人视觉系统提供有效解决方案。
Abstract:
To address the challenges of low efficiency and unstable quality associated with manual identification methods in industrial bisporus mushroom cultivation, an enhanced SIAF-YOLOv8 algorithm for bisporus mushroom identification was developed to achieve automatic and precise detection. First, the feature fusion architecture was restructured using a four-layer asymptotic feature pyramid network (AFPN), thereby reducing the semantic gap through multi-level feature interaction. Second, the SimAM 3D attention mechanism was integrated into the backbone network to enable adaptive parametric feature weighting. A self-built dataset comprising 2 111 images and 60 053 samples of Agaricus bisporus was utilized for validation. Results show that the improved algorithm achieved AP50 and AP50-95 scores of 98.5% and 85.7%, respectively, representing improvement of 1.0% and 4.6% over the original YOLOv8n. Additionally, the model size was reduced to 5.6 MB, with a single-image inference time of 12ms. Compared to several contemporary general-purpose object detection algorithms, SIAF-YOLOv8 offers an optimal balance between detection accuracy, speed, and model compactness, providing an effective solution for agricultural robot vision systems.

参考文献/References:

［1］ LI C T，XU S. Edible mushroom industry in China：current state and perspectives［J］. Applied Microbiology and Biotechnology，2022，106（11）：3949-3955.［2］ 刘雪，莫桐桐，郑文瑞，等. 双孢菇生产智能控制技术研究现状及发展趋势［J］. 智慧农业导刊，2023，3（2）：7-12.［3］ 王磊磊，王斌，李东晓，等. 基于改进YOLOv5的菇房平菇目标检测与分类研究［J］. 农业工程学报，2023，39（17）：163-171.［4］ 华金. 基于机器视觉的双孢菇品质检测与分级方法研究［D］. 淄博：山东理工大学，2024. ［5］ 韩瑞青. 基于深度视觉的成簇双孢菇采摘方法研究［D］. 哈尔滨：哈尔滨工业大学，2023.［6］ 吕强，林刚，蒋杰，等. 基于改进YOLOv5s模型的自然场景中绿色柑橘果实检测［J］. 农业工程学报，2024，40（18）：147-154.［7］ 杨震宇，汪小旵，祁子涵，等. 基于改进YOLOv8的草莓识别与果梗采摘关键点检测［J］. 农业工程学报，2024，40（18）：167-175.［8］ REN S Q，HE K M，GIRSHICK R，et al. Faster R-CNN：towards real-time object detection with region proposal networks［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence，2017，39（6）：1137-1149.［9］ REDMON J，FARHADI A. YOLOv3：an incremental improvement［EB/OL］. (2018-04-08)［2025-01-02］. https:∥arxiv.org/obs/1804.02767. ［10］ LI C Y，LI L L，JIANG H L，et al. YOLOv6：a single-stage object detection framework for industrial applications［EB/OL］.(2022-09-07)［2025-01-02］.https:∥arxiv.org/abs/2209.02976.［11］ WANG A，CHEN H，LIU L H，et al. YOLOv10：real-time end-to-end object detection［EB/OL］.(2024-05-23)［2025-01-10］.https:∥arxiv.org/abs/2405.14458.

更新日期/Last Update: 2025-12-25