YOLO系列模型在水果缺陷检测的研究进展及发展趋势

doi:10.12133/j.smartag.SA202603017

Smart Agriculture

• •

YOLO系列模型在水果缺陷检测的研究进展及发展趋势

修贤超¹, 费士祺¹^,²^,³, 张伟¹(), 黄文倩²^,³, 苗中华¹

^1. 上海大学机电工程与自动化学院，上海 200444，中国
^2. 北京农林科学院，北京 100097，中国
^3. 北京市农林科学院信息技术研究中心，北京 100097，中国

收稿日期:2026-03-13 出版日期:2026-05-22
基金项目:
国家自然科学基金(32401712); 上海市自然科学基金(24ZR1424800); 博士后科学基金(2024M751937)
作者简介:
修贤超，副教授，硕士生导师，研究方向为大模型与具身智能。E-mail：xcxiu@shu.edu.cn
通信作者:
张伟，博士，助理研究员，研究方向为农业机器人具身智能。E-mail：wzhang23@shu.edu.cn

Research Progress and Development Trends of YOLO Family of Algorithms in Fruit Defect Detection

XIU Xianchao¹, FEI Shiqi¹^,²^,³, ZHANG Wei¹(), HUANG Wenqian²^,³, MIAO Zhonghua¹

^1. School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, China
^2. Intelligent Equipment Research Center, Beijing Academy of Agriculture and Forestry Sciences, Beijing 100097, China
^3. Information Technology Research Center, Beijing Academy of Agriculture and Forestry Sciences, Beijing 100097, China

Received:2026-03-13 Online:2026-05-22
Foundation items:National Natural Science Foundation of China(32401712); Natural Science Foundation of Shanghai(24ZR1424800); China Postdoctoral Science Foundation(2024M751937)
About author:
XIU Xianchao, E-mail: xcxiu@shu.edu.cn
Corresponding author:
ZHANG Wei, E-mail: wzhang23@shu.edu.cn

摘要/Abstract

摘要：

【目的/意义】 作为水果产后的关键环节，高效精准的质量检测对保障食品安全与提升产业效率至关重要。本文综述了YOLO模型在水果缺陷检测领域的研究进展，旨在为农业水果缺陷检测技术的进一步发展提供理论支撑与技术参考。［进展］近年来，基于深度学习的目标检测算法，特别是YOLO（You Only Look Once）系列模型在实时性、高精度和低计算量的优势，在水果缺陷检测研究获得了出色的效果。本文聚焦水果缺陷检测，首先回顾了YOLOv1至YOLO26的发展历程，并分析每代YOLO的创新改进。在此基础上，综述了YOLO系列模型在多种水果缺陷检测任务中的应用现状，列举了这些算法在应对水果缺陷检测挑战方面的贡献，具体包括内部与隐性缺陷的识别、复杂表面形态下的缺陷检测，以及模型的轻量化与边缘部署。同时，本文也探讨了这些技术如何满足高速生产线的实时处理，以及在资源受限的硬件设备上实现高效部署的工业需求。 【结论/展望】 最后，总结了算法局限性，数据集特异性，硬件部署约束和复杂检测环境影响四方面的挑战，并从公共数据集的构建，模型可解释性，人机协作策略机制和大模型与YOLO的结合的四个方面展望了未来研究方向。

关键词: 深度学习, 目标检测, YOLO, 自动化, 水果缺陷检测

Abstract:

[Significance] Efficient and accurate quality detection is a key link in the post-harvest fruit industry. It is critical for ensuring food safety and improving industrial efficiency. With the growth of global population and the increasing demand for food safety, traditional manual inspection has become incompetent. Manual inspection relies heavily on the experience of operators. It is time-consuming, labor-intensive and subjective. The results are often unstable and inaccurate. It also may cause secondary damage to fruits. In addition, the rising cost of human resources makes the transformation of the fruit industry towards automation and intelligence an inevitable trend. Computer vision and machine learning technologies provide a solution. They can improve detection accuracy and efficiency, while reducing labor costs. Among various object detection algorithms, the YOLO (You Only Look Once) series stands out for its balance of real-time performance, high accuracy and low computational cost. It has achieved excellent results in fruit defect detection. However, there is still a lack of comprehensive reviews on the application of YOLO series algorithms in this field. The purpose of this paper is to fill the blank of YOLO series algorithms in the field of fruit defect detection and provide strong support for the further development of agricultural fruit defect detection technology. [Progress] Focusing on fruit defect detection, this paper reviews the development history of YOLO models from YOLOv1 to YOLO26. It also analyzes the innovative improvements of each generation. The development of YOLO models can be divided into three stages. The first stage (YOLOv1-YOLOv3) established the single-stage detection framework. YOLOv1 proposed the idea of single-shot detection, laying a preliminary foundation for real-time fruit defect detection. YOLOv2 introduced the anchor box mechanism and a more efficient backbone network, improving the positioning robustness for fruits and their defects of different sizes and shapes. YOLOv3 proposed a multi-scale feature pyramid, enhancing the detection ability for small defects such as early lesions and minor bruises. The second stage (YOLOv4-YOLOv8) focused on modular design, engineering construction and efficiency optimization. YOLOv4 clarified the modular design of backbone, neck and head, providing a clear framework for algorithm improvement. YOLOv5, YOLOv6 and YOLOv8 made continuous breakthroughs in lightweight design, anchor-free detection and decoupled head structure. YOLOv7 introduced the E-ELAN (Extended Efficient Layer Aggregation Network) architecture and auxiliary head, balancing detection accuracy and speed. The third stage (YOLOv9-YOLO26) focused on information extraction and transmission, post-processing simplification and high-level semantic modeling. YOLOv9 alleviated information attenuation in deep networks. YOLOv10 simplified post-processing to meet the high throughput needs of online sorting systems. YOLOv13 and YOLO26 further improved the model's performance in complex scenarios. This review also summarizes the application of YOLO series models in various fruit defect detection tasks, including apples, citrus, tomatoes and pears. It focuses on three key challenges: detection of internal and latent defects, defect detection under complex surface morphologies, and model lightweighting and edge deployment. For internal and latent defects, researchers combine YOLO models with multi-modal imaging technologies such as near-infrared and hyperspectral imaging. For complex surface morphologies, they improve the model's feature extraction ability by introducing attention mechanisms and multi-scale feature fusion. For lightweight and edge deployment, they adopt strategies such as lightweight architecture design, model compression and training optimization. [Conclusions and Prospects] The YOLO series models have become one of the most widely used technical frameworks in fruit defect detection. They have achieved remarkable phased results. However, they still face several challenges. These include algorithm limitations, such as the coupling optimization conflict between positioning and classification, and the limitation of local feature extraction. Dataset specificity leads to poor model generalization. Hardware constraints restrict the deployment of models on edge devices. Complex detection environments reduce the model's robustness. Future research directions include four aspects. First, build standardized and large-scale public fruit defect datasets, covering different fruit varieties and defect types. Second, improve the interpretability of YOLO models, building intuitive explanation interfaces for non-professional users. Third, develop human-machine collaboration strategies, combining the advantages of deep learning models and farmers' practical experience. Fourth, integrate YOLO with large models, optimizing data generation strategies and model lightweighting to achieve efficient deployment. This review aims to provide theoretical support and technical reference for the further development of agricultural fruit defect detection technology.

Key words: deep learning, object detection, YOLO, automation, fruit defect detection

中图分类号:

S126
TP29

修贤超, 费士祺, 张伟, 黄文倩, 苗中华. YOLO系列模型在水果缺陷检测的研究进展及发展趋势[J]. 智慧农业(中英文), doi: 10.12133/j.smartag.SA202603017.

XIU Xianchao, FEI Shiqi, ZHANG Wei, HUANG Wenqian, MIAO Zhonghua. Research Progress and Development Trends of YOLO Family of Algorithms in Fruit Defect Detection[J]. Smart Agriculture, doi: 10.12133/j.smartag.SA202603017.

图/表 8

表1

图1

图2

表2

图3

表3

表4

基于YOLO系列模型的复杂表面环境下水果缺陷检测相关研究成果

年份	模型	水果	改进方法	研究难点	mAP_0.5/%
2022	YOLOv8^［60］	苹果	1）引入空间深度卷积（Space-to-Depth Convolution， SPD-Conv） 2）引入多尺度空洞注意力机制（Multi-scale Empty Attention， MSDA） 3）引入上下文引导特征金字塔网络（Context Guided Feature Pyramid Network， CGFPN）	表面坑凹区域与缺陷相似易误检	91.4
2025	YOLOv8^［61］	苹果	1）引入部分卷积（Partial Convolution，PConv）模块 2）高效多尺度注意力（Efficient Multi-scale Attention，EMA）机制 3）最小点距离交并比（Minimum Point Distance Intersection over Union） 4）引入Slim-neck架构	表面坑凹区域与缺陷相似易误检	90.1
2023	YOLOv5^［62］	苹果	1）引入TRANS（Transformer）模块 2）引入BiFPN 3）基于归一化的注意力模块（Normalization-based Attention Module，NAM）	表面坑凹区域与缺陷相似易误检	98.9
2024	YOLOv5^［63］	梨	1）BAM（Bottleneck Attention Module）注意力机制 2）设计C3-GDC（C3-GhostDynamicConv）卷积模块 3）明智交并比（Wise Intersection over Union，WIoU）损失函数	花萼/果梗与缺陷相似易误检	93.9
2025	YOLOv5^［64］	梨	1）基于双层路由注意力机制的视觉 Transformer（Bi-Level Routing Vision Transformer，Bi Former） 2）引入双向特征金字塔网络（Bidirectional Feature Pyramid Network，BiFPN）	花萼/果梗与缺陷相似易误检	68.2
2023	YOLOv10^［65］	柑橘	1）引入线性可变形卷积（Linear Deformable Convolution，LDConv） 2）引入广义特征金字塔网络（Generalized Feature Pyramid Network，GFPN）	缺陷微小且复杂多变易漏检	98.7
2024	YOLOv5^［66］	柑橘	1）三维坐标注意力（Three-dimensional Coordinate Attention，TDCA）机制 2）引入CT（Contextual Transformer）模块 3）引入BiFPN	缺陷微小且复杂多变易漏检	98.7
2024	YOLOv8^［67］	柠檬	1）CBAM注意力机制 2）设计C2f-SAS（Switchable Atrous Convolution）模块	缺陷形态复杂多变易漏检	90.62
2024	YOLOv5^［68］	苹果	1）设计DSCS卷积模块 2）指数线性单元（Exponential Linear Unit，ELU）激活函数 3）WIoU损失函数	缺陷微小且复杂多变易漏检	90.9
2024	YOLOv7^［69］	苹果	1）CBAM和ECA注意力机制 2）引入深度可分离卷积策略 3）参数化修正线性单元（Parametric Rectified Linear Unit，PReLU）损失函数	缺陷微小且复杂多变易漏检	90.7
2026	YOLOv11^［70］	苹果	1）设计三视角成像系统 2）引入非局部自注意力残差多层感知机模块（Non-Local Attention Residual Multi-layer perceptron，NARM）	缺陷微小且复杂多变易漏检	89.8
2024	YOLOv8^［71］	李子	1）引入SPD-Conv模块 2）引入扩张式残差分割网络（Dilation-wise Residual Segmentation，DWRSeg）模块 3）高效交并比（Efficient Intersection over Union，EIoU）损失函数	缺陷微小且复杂多变易漏检	87.8
2025	YOLOv11^［72］	番茄	1）引入小波深度可分离卷积（ Wavelet Depth Separable Convolution，WDWConv） 2）设计WC3k2模块	不规则缺陷易漏检	88
2024	YOLOv7^［73］	猕猴桃	1）CBAM注意力机制 2）设计C3-Ghost模块 3）优化损失函数	不规则缺陷易漏检	93
2024	YOLOv7^［74］	红枣	1）简单注意力机制（Simple Attention Module，SimAM） 2）引入深度可分离卷积（Depthwise Separable Convolution，DSConv） 3）引入ByteTrack跟踪算法	密集排列遮挡且不规则缺陷易漏检	85.3
2025	YOLOv5^［75］	红枣	1）坐标注意力（Coordinate Attention，CA）机制 2）增加小目标检测头	密集排列遮挡且不规则缺陷易漏检	91.6
2025	YOLOv7^［76］	葡萄	1）WIoU损失函数 2）Bi-Former注意力机制	果实小且密集排列遮挡易漏检	80.1
2023	YOLOX^［77］	樱桃	1）CBAM注意力机制 2）改进损失函数	果实小且密集排列遮挡易漏检	91.84
2025	YOLOv8^［78］	樱桃	1）引入DySlimNeck动态颈部层 2）引入Detect-FRM（Detect-Fourier Representation Module）检测头 3）引入FasterNet-RepMixer网络	果实小且密集排列遮挡易漏检	91.5

表4

表5

基于YOLO系列算法的水果缺陷检测模型轻量化相关研究成果

年份	模型	水果	改进方法	研究难点	mAP_0.5/%
2025	YOLOv11^［79］	梨	1）引入BiFPN 2）CBAM注意力机制 3）引入TR2（Two Transformer）模块 4）双层迁移学习	实际场景计算资源受限模型难以部署	92.7
2023	YOLOv4^［80］	梨	1）PConv模块 2）EMA机制 3）MPDIoU损失函数 4）引入Slim-neck架构轻量化颈部网络	实际场景计算资源受限模型难以部署	85.52
2025	YOLOv5^［81］	苹果	1）多尺度检测头简化为双尺度或单尺度 2）移除冗余卷积层	实际场景计算资源受限模型难以部署	99.8
2024	YOLOv5^［82］	苹果	1）引入LightweightC3模块作为骨干网络的基本构建模块 2）设计EnhancedC3模块替代模型特征融合部分的C3模块 3）引入GSConv模块	实际场景计算资源受限模型难以部署	96.12
2022	YOLOv5^［83］	芒果	1）引入PPLCNet网络 2）引入GhostNet网络	实际场景计算资源受限模型难以部署	94.3
2025	YOLOv5^［84］	百香果	1）设计了StarC3SE模块替代C3模块 2）CBAM注意力机制 3）距离交并比（Distance Intersection over Union， DIoU）损失函数	实际场景计算资源受限模型难以部署	88.4
2024	YOLOv5^［85］	苹果	1）引入EfficientNetV模块 2）EMA机制	实际场景计算资源受限模型难以部署	93.6
2023	YOLOv4^［86］	柑橘	1）对网络进行剪枝优化移除冗余通道 2）去除冗余预测框	实际场景计算资源受限模型难以部署	94.56
2022	YOLOv4^［87］	苹果	1）与BiSeNet V2语义网络结合构成了双模型级联检测框架 2）对网络进行剪枝优化移除冗余通道	实际场景计算资源受限模型难以部署	92.42
2025	YOLOv12^［88］	番茄	1）引入RFEM模块 2）设计Dysample-Slim-Neck机制 3）引入动态卷积替代检测头中的传统卷积结构	实际场景计算资源受限模型难以部署	88.5
2024	YOLOv8^［89］	苹果	知识蒸馏	实际场景计算资源受限模型难以部署	97.8
2025	YOLOv8^［90］	百香果	1）设计MFSO（More Focus on Small Objects）网络结构 2）动态重参数化（Dynamic Reparameterization， DyRep） 3）NWD（Normalized Wasserstein Distance）损失函数 4）模型剪枝	实际场景计算资源受限模型难以部署	94.6
2024	YOLOv3^［91］	莲蓬	1）先验锚框尺度调整 2）引入金字塔池化层（Spatial Pyramid Pooling， SPP） 3）使用遗传算法改进网络超参数	实际场景计算资源受限模型难以部署	96.8

表5

参考文献 98

[1]	PANDISELVAM R, SUBHASHINI S, BANUU PRIYA E P, et al. Ozone based food preservation: a promising green technology for enhanced food safety[J]. Ozone: Science & Engineering, 2019, 41(1): 17-34.
[2]	PANDISELVAM R, SUNOJ S, MANIKANTAN M R, et al. Application and kinetics of ozone in food preservation[J]. Ozone: Science & Engineering, 2017, 39(2): 115-126.
[3]	RAWAT W, WANG Z H. Deep convolutional neural networks for image classification: A comprehensive review[J]. Neural Computation, 2017, 29(9): 2352-2449.
[4]	田有文, 吴伟, 卢时铅, 等. 深度学习在水果品质检测与分级分类中的应用[J]. 食品科学, 2021, 42(19): 260-270.
	TIAN Y W, WU W, LU S Q, et al. Application of deep learning in fruit quality detection and grading[J]. Food Science, 2021, 42(19): 260-270.
[5]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2016: 779-788.
[6]	宋怀波, 尚钰莹, 何东健. 果实目标深度学习识别技术研究进展[J]. 农业机械学报, 2023, 54(1): 1-19.
	SONG H B, SHANG Y Y, HE D J. Review on deep learning technology for fruit target recognition[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(1): 1-19.
[7]	WANG H G, MEI M W, LI J B. Research progress on non-destructive detection of internal quality of fruits with large size and thick peel: A review[J]. Agriculture, 2023, 13(9): 1838.
[8]	SOLTANI FIROUZ M, SARDARI H. Defect detection in fruit and vegetables by using machine vision systems and image processing[J]. Food Engineering Reviews, 2022, 14(3): 353-379.
[9]	GAIKWAD A, DESHPANDE M, BHOLE V. Fruits defect detection: a review with image processing techniques[M]// Responsible Production and Consumption. LondonCRC Press 2024: 221-228.
[10]	张子申, 程红, 耿文娟, 等. 高光谱成像技术在水果品质检测中的应用研究进展[J]. 智慧农业(中英文), 2025, 7(5): 52-66.
	ZHANG Z S, CHENG H, GENG W J, et al. Research advances in hyperspectral imaging technology for fruit quality assessment[J]. Smart Agriculture, 2025, 7(5): 52-66.
[11]	徐文青, 丁爱霞, 朱文卿. 基于数字图像处理技术的水果分级检测研究进展[J]. 食品安全导刊, 2025(4): 183-189.
	XU W Q, DING A X, ZHU W Q. Research progress of fruit grading detection based on digital image processing technology[J]. China Food Safety Magazine, 2025(4): 183-189.
[12]	JOHN J, ANSARI R, DAFEDAR S, et al. Recent developments in fruit quality detection using deep learning: A review[C]// Proceedings of International Conference on Information Technology and Artificial Intelligence. Cham, Germany: Springer, 2026: 285-299.
[13]	麻剑钧, 刘晓慈, 金龙新, 等. 基于机器视觉的农作物病害识别研究进展[J]. 湖南农业科学, 2023(9): 97-100.
	MA J J, LIU X C, JIN L X, et al. Research progress of crop disease recognition based on machine vision[J]. Hunan Agricultural Sciences, 2023(9): 97-100.
[14]	周长建, 宋佳, 向文胜. 基于人工智能的作物病害识别研究进展[J]. 植物保护学报, 2022, 49(1): 316-324.
	ZHOU C J, SONG J, XIANG W S. Research progresses in artificial intelligence-based crop disease identification[J]. Journal of Plant Protection, 2022, 49(1): 316-324.
[15]	郭文娟, 冯全, 李相周. 基于农作物病害检测与识别的卷积神经网络模型研究进展[J]. 中国农机化学报, 2022, 43(10): 157-166.
	GUO W J, FENG Q, LI X Z. Research progress of convolutional neural network model based on crop disease detection and recognition[J]. Journal of Chinese Agricultural Mechanization, 2022, 43(10): 157-166.
[16]	IYYER M, MANJUNATHA V, BOYD-GRABER J, et al. Deep unordered composition rivals syntactic methods for text classification[C]// Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (volume 1: Long Papers). Beijing, China: Association for Computational Linguistics, 2015: 1681-1691.
[17]	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]// Communications of the ACM. ACM, 2017: 84-90.
[18]	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]// Proceedings of the International Conference on Learning Representations. San Diego, USA: ICLR, 2015.
[19]	SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2015: 1-9.
[20]	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2016: 770-778.
[21]	TAN M, LE Q V. EfficientNet: Rethinking model scaling for convolutional neural networks[C]// International Conference on Machine Learning. New York, USA: PMLR, 2019: 6105-6114.
[22]	HOWARD A, SANDLER M, CHEN B, et al. Searching for MobileNetV3[C]// 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2019: 1314-1324.
[23]	LIU Z, MAO H Z, WU C Y, et al. A ConvNet for the 2020s[C]// 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2022: 11966-11976.
[24]	DING X, ZHANG X, MA N, et al. Scaling up your kernels to 31×31: Revisiting large kernel design in CNNs[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2022: 11021-11030.
[25]	LIN X, YAN Z Q, DENG X B, et al. ConvFormer: plug-and-play CNN-style transformers forImproving medical image segmentation[C]// Medical Image Computing and Computer Assisted Intervention. Cham, Germany: Springer, 2023: 642-651.
[26]	TAN M, LE Q V. EfficientNetV2: Smaller models and faster training[C]// International Conference on Machine Learning. New York, USA: PMLR, 2021: 10096-10106.
[27]	LIM S N, LU J W, RAO Y M, et al. HorNet: efficient high-order spatial interactions with recursive gated convolutions[C]// Advances in Neural Information Processing Systems 35. New Orleans, Louisiana, USA. Neural Information Processing Systems Foundation, Inc. (NeurIPS), 2022: 10353-10366.
[28]	GIRSHICK R. Fast R-CNN[C]// 2015 IEEE International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2015: 1440-1448.
[29]	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.
[30]	LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2017: 936-944.
[31]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]// Computer Vision – ECCV 2016. Cham, Germany: Springer, 2016: 21-37.
[32]	SUN C, AI Y B, WANG S, et al. Dense-RefineDet for traffic sign detection and classification[J]. Sensors, 2020, 20(22): 6570.
[33]	FU C Y, LIU W, RANGA A, et al. DSSD: deconvolutional single shot detector[EB/OL]. arXiv: 1701.06659, 2017.
[34]	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]// 2017 IEEE International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2017: 2999-3007.
[35]	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2014: 580-587.
[36]	LIAO S Y, WANG J B, YU R Y, et al. CNN for situations understanding based on sentiment analysis of twitter data[J]. Procedia Computer Science, 2017, 111: 376-381.
[37]	YEERJIANG A, WANG Z, HUANG X, et al. YOLOv1 to YOLOv10: A comprehensive review of YOLO variants and their application in medical image detection[J]. Journal of Artificial Intelligence Practice, 2024, 7(3): 112–122.
[38]	RABBANI ALIF MAL, HUSSAIN M. YOLOv1 to YOLOv10: a comprehensive review of YOLO variants and their application in the agricultural domain[EB/OL]. arXiv: 2406.10139, 2024.
[39]	HUSSAIN M. YOLO-v1 to YOLO-v8, the rise of YOLO and its complementary nature toward digital manufacturing and industrial defect detection[J]. Machines, 2023, 11(7): 677.
[40]	REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2017: 6517-6525.
[41]	FACHKHA C, DEBBABI M. Darknet as a source of cyber intelligence: survey, taxonomy, and characterization[J]. IEEE Communications Surveys & Tutorials, 2016, 18(2): 1197-1227.
[42]	REDMON J, FARHADI A, et al. YOLOv3: An incremental improvement[EB/OL]// arXiv:1804.02767v1, 2018.
[43]	BOCHKOVSKIY A, WANG C Y, LIAO H M. YOLOv4: Optimal speed and accuracy of object detection[EB/OL]. arXiv: 2004.10934, 2020.
[44]	WANG C Y, MARK LIAO H Y, WU Y H, et al. CSPNet: a new backbone that can enhance learning capability of CNN[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Piscataway, New Jersey, USA: IEEE, 2020: 1571-1580.
[45]	MA Z, LI M, WANG Y G. PAN: path integral based convolution for deep graph neural networks[EB/OL]. arXiv: 1904.10996, 2019.
[46]	DING X H, ZHANG X Y, MA N N, et al. RepVGG: Making VGG-style ConvNets great again[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2021: 13728-13737.
[47]	WANG C Y, BOCHKOVSKIY A, LIAO H M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2023: 7464-7475.
[48]	WANG C Y, YEH I H, LIAO H M. YOLOv9: Learning what you want to learn using programmable gradient information[EB/OL]. arXiv: 2402.13616, 2024.
[49]	CHEN H, CHEN K, DING G G, et al. YOLOv10: Real-time end-to-end object detection[C]// Advances in Neural Information Processing Systems 37. Vancouver, BC, Canada. Neural Information Processing Systems Foundation, Inc. (NeurIPS), 2024: 107984-108011.
[50]	TIAN Y J, YE Q X, DOERMANN D. YOLOv12: attention-centric real-time object detectors[EB/OL]. arXiv: 2502.12524, 2025.
[51]	LEI M Q, LI S Q, WU Y H, et al. YOLOv13: Real-time object detection with hypergraph-enhanced adaptive visual perception[EB/OL].: arXiv: 2506.17733, 2025.
[52]	CAI Z L, SUN C J, ZHANG H L, et al. Developing universal classification models for the detection of early decayed Citrus by structured-illumination reflectance imaging coupling with deep learning methods[J]. Postharvest Biology and Technology, 2024, 210: 112788.
[53]	FAN S X, LIANG X T, HUANG W Q, et al. Real-time defects detection for apple sorting using NIR cameras with pruning-based YOLOV4 network[J]. Computers and Electronics in Agriculture, 2022, 193: 106715.
[54]	LU Y Z, GONG M L, LI J, et al. Strawberry defect identification using deep learning infrared–visible image fusion[J]. Agronomy, 2023, 13(9): 2217.
[55]	HU W X, XIONG J T, LIANG J H, et al. A method of Citrus epidermis defects detection based on an improved YOLOv5[J]. Biosystems Engineering, 2023, 227: 19-35.
[56]	WANG D Y, ZHANG M, ZHU Q B, et al. Beyond human vision: highlighting and detecting inconspicuous defects in Citrus fruits using UV-fluorescence imaging and YOLO11[J]. Postharvest Biology and Technology, 2026, 235: 114190.
[57]	WANG L X, WANG Z S, ZHAO X Y, et al. Enhanced research on YOLOv12 detection of apple defects by integrating filter imaging and color space reconstruction[J]. Electronics, 2025, 14(21): 4259.
[58]	YUAN S Q, CAO Y, CHENG X. Research on strawberry quality grading based on object detection and stacking fusion model[J]. IEEE Access, 2023, 11: 137475-137484.
[59]	朱文铭. 基于深度学习的草莓缺陷检测算法研究[D]. 湛江: 广东海洋大学, 2023.
	ZHU W M. Research ON STRAWBERRY DEFECT DETECTION ALGORITHM BASED ON DEEP LEarning [D]. Zhanjiang: Guangdong Ocean University, 2023.
[60]	SHAN W, YUE Y R. Apple defect detection in complex environments[J]. Electronics, 2024, 13(23): 4844.
[61]	王会征, 李新龙, 薄萍, 等. 基于改进YOLOv8n的自然场景下苹果外观品质检测方法[J]. 农业工程学报, 2025, 41(11): 173-182.
	WANG H Z, LI X L, BO P, et al. Detecting apple appearance quality under natural scenes using improved YOLOv8n[J]. Transactions of the Chinese Society of Agricultural Engineering, 2025, 41(11): 173-182.
[62]	张嘉琪, 徐啟蕾. 基于NAM-YOLO网络的苹果缺陷检测算法[J]. 计算机与现代化, 2023(10): 53-58, 64.
	ZHANG J Q, XU Q L. Apple defect detection algorithm based on NAM-YOLO network[J]. Computer and Modernization, 2023(10): 53-58, 64.
[63]	WANG B Y, HUA J, XIA L M, et al. A defect detection method for Akidzuki pears based on computer vision and deep learning[J]. Postharvest Biology and Technology, 2024, 218: 113157.
[64]	陈峻生. 基于深度学习的梨缺陷检测与分级[D]. 福州: 福建农林大学, 2025.
	CHEN J S. Research on defect detection and grading of pears based on deep learning [D]. Fuzhou: Fujian Agriculture and Forestry University, 2025.
[65]	WANG L, YE R, CHEN Y Q, et al. YOLOv10-LGDA: An improved algorithm for defect detection in Citrus Fruits across diverse backgrounds[J]. Plants, 2025, 14(13): 1990.
[66]	LU J Q, CHEN W D, LAN Y B, et al. Design of Citrus peel defect and fruit morphology detection method based on machine vision[J]. Computers and Electronics in Agriculture, 2024, 219: 108721.
[67]	WANG S Y, LI Q J, YANG T, et al. LSD-YOLO: enhanced YOLOv8n algorithm for efficient detection of lemon surface diseases[J]. Plants, 2024, 13(15): 2069.
[68]	朱琦, 周德强, 盛卫锋, 等. 基于DSCS-YOLO的苹果表面缺陷检测方法[J]. 南京农业大学学报, 2024, 47(3): 592-601.
	ZHU Q, ZHOU D Q, SHENG W F, et al. Apple surface defect detection method based on DSCS-YOLO[J]. Journal of Nanjing Agricultural University, 2024, 47(3): 592-601.
[69]	徐江鹏, 王传安. 基于轻量化改进YOLO v7-Tiny算法的苹果检测与分类[J]. 江苏农业科学, 2024, 52(23): 221-229.
	XU J P, WANG C A. Detection and classification of apple based on lightweight improved YOLO v7-Tiny algorithm[J]. Jiangsu Agricultural Sciences, 2024, 52(23): 221-229.
[70]	孙国祥, 陈光宇, 汪小旵, 等. 基于三视角成像的苹果表面缺陷检测方法[J]. 农业工程学报, 2026, 42(3): 348-358.
	SUN G X, CHEN G Y, WANG X C, et al. Three-view imaging-based method for defect detection on apple surfaces[J]. Transactions of the Chinese Society of Agricultural Engineering, 2026, 42(3): 348-358.
[71]	霍新梅. 自然环境中李子果实成熟度与表面缺陷检测方法研究[D]. 合肥: 安徽农业大学, 2024.
	HUO X M. Research on maturity and surface defect detection method of plum fruit in natural environment [D]. Hefei: Anhui Agricultural University, 2024.
[72]	朱婷婷, 滕广, 张亚军, 等. 基于改进YOLOv11的番茄表面缺陷检测方法[J]. 农业机械学报, 2025, 56(6): 546-553.
	ZHU T T, TENG G, ZHANG Y J, et al. Improved YOLOv11 method for surface defect detection of tomato[J]. Transactions of the Chinese Society for Agricultural Machinery, 2025, 56(6): 546-553.
[73]	许川. 猕猴桃表面缺陷智能检测与分级方法研究[D]. 柳州: 广西科技大学, 2024.
	XU C. Research on intelligent detection and grading method for surface defects of kiwifruit [D]. Liuzhou: Guangxi University of Science and Technology, 2024.
[74]	刘鑫, 马本学, 李玉洁, 等. 基于改进YOLOv7-ByteTrack的干制哈密大枣缺陷检测与计数系统[J]. 农业工程学报, 2024, 40(3): 303-312.
	LIU X, MA B X, LI Y J, et al. Detecting and counting defects in dried Hami jujube using improved YOLOv7-ByteTrack[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(3): 303-312.
[75]	史鹏涛, 田政伟, 李晓泽, 等. 基于改进YOLO v5s算法的红枣缺陷检测与分拣方法[J]. 江苏农业科学, 2025, 53(5): 83-88.
	SHI P T, TIAN Z W, LI X Z, et al. Defect detection and sorting method of jujube based on improved YOLO v5s algorithm[J]. Jiangsu Agricultural Sciences, 2025, 53(5): 83-88.
[76]	LI C Y, ZHANG S C, MA Z J. RF-YOLOv7: A model for the detection of poor-quality grapes in natural environments[J]. Agriculture, 2025, 15(4): 387.
[77]	刘敬宇, 裴悦琨, 常志远, 等. 基于改进YOLOX模型的樱桃缺陷及分级检测[J]. 食品与机械, 2023, 39(1): 139-145.
	LIU J Y, PEI Y K, CHANG Z Y, et al. Cherry defect and classification detection based on improved YOLOX model[J]. Food & Machinery, 2023, 39(1): 139-145.
[78]	LUAN F, FAN K L, XU X H, et al. Cherry-YOLO: enhanced real-time detection of Cherry ripeness and defects with optimised YOLOv8[J]. Computing, 2025, 107(10): 198.
[79]	李成轩, 赵春江, 张驰, 等. 基于AIC-YOLOv11n模型的砀山梨多表面缺陷检测方法[J]. 农业工程学报, 2025, 41(14): 320-328.
	LI C X, ZHAO C J, ZHANG C, et al. Multi-surface defect detection method for Dangshan pears based on AIC-YOLOv11n model[J]. Transactions of the Chinese Society of Agricultural Engineering, 2025, 41(14): 320-328.
[80]	XIE Y H, ZHONG X Y, ZHAN J L, et al. ECLPOD: An extremely compressed lightweight model for pear object detection in smart agriculture[J]. Agronomy, 2023, 13(7): 1891.
[81]	OLGUÍN-ROJAS J C, VASQUEZ J I, DE JESÚS LÓPEZ-CANTEÑS G, et al. A lightweight YOLO-based architecture for apple detection on embedded systems[J]. Agriculture, 2025, 15(8): 838.
[82]	鄢紫, 陈良艳, 刘卫华, 等. 基于YOLO-FFD的水果品种和新鲜度识别方法[J]. 食品与机械, 2024, 40(1): 115-121.
	YAN Z, CHEN L Y, LIU W H, et al. Fruit variety and freshness recognition method based on YOLO-FFD[J]. Food & Machinery, 2024, 40(1): 115-121.
[83]	聂衍文, 徐建. 基于YOLOv5s的移动端轻量化芒果果面缺陷检测应用设计[J]. 武汉轻工大学学报, 2022, 41(5): 89-95, 103.
	NIE Y W, XU J. Mobile terminal mango defect detection application design based on lightweight YOLOv5s[J]. Journal of Wuhan Polytechnic University, 2022, 41(5): 89-95, 103.
[84]	ZHOU Y, LI Z Y, XUE S, et al. Lightweight SCD-YOLOv5s: The detection of small defects on passion fruit with improved YOLOv5s[J]. Agriculture, 2025, 15(10): 1111.
[85]	LV L J, YILIHAMU Y, YE Y L. Apple surface defect detection based on lightweight improved YOLOv5s[C]// International Journal of Information and Communication Technology. New York, USA: ACM, 2024: 113-128.
[86]	LIANG X T, JIA X Y, HUANG W Q, et al. Real-time grading of defect apples using semantic segmentation combination with a pruned YOLO V4 network[J]. Foods, 2022, 11(19): 3150.
[87]	梁晓婷, 庞琦, 杨一, 等. 基于YOLOv4模型剪枝的番茄缺陷在线检测[J]. 农业工程学报, 2022, 38(6): 283-292.
	LIANG X T, PANG Q, YANG Y, et al. Online detection of tomato defects based on YOLOv4 model pruning[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(6): 283-292.
[88]	LIANG Z H, ZHU T T, TENG G, et al. YOLO-RGDD: A novel method for the online detection of tomato surface defects[J]. Foods, 2025, 14(14): 2513.
[89]	方晨羽. 基于深度学习的果实采前识别与采后品质检测研究[D]. 杭州: 浙江科技大学, 2024.
	FANG C Y. Research on pre-harvest recognition and post-harvest quality detection of fruits based on deep learning [D]. Hangzhou: Zhejiang University of Science and Technology, 2024.
[90]	CHEN D J, LIN F, LU C H, et al. YOLOv8-MDN-Tiny: A lightweight model for multi-scale disease detection of postharvest golden passion fruit[J]. Postharvest Biology and Technology, 2025, 219: 113281.
[91]	张雷, 严昊, 贾永镒, 等. 采用改进YOLOv3-Tiny模型的轻量化莲蓬质量分级算法[J]. 农业工程学报, 2024, 40(23): 248-257.
	ZHANG L, YAN H, JIA Y Y, et al. Lightweight classification of Lotus seedpod quality using improved YOLOv3-Tiny model[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(23): 248-257.
[92]	ZHANG J N, LI X T, LI J, et al. Rethinking mobile block for efficient attention-based models[C]// 2023 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2023: 1389-1400.
[93]	梁秀满, 肖寒. 基于SDD-YOLO的轻量级带钢缺陷实时检测算法[J]. 中国测试, 2025, 51(3): 154-161.
	LIANG X M, XIAO H. Lightweight strip defect real-time detection algorithm based on SDD-YOLO[J]. China Measurement & Test, 2025, 51(3): 154-161.
[94]	YU Z P, HUANG H B, CHEN W J, et al. YOLO-FaceV2: : A scale and occlusion aware face detector[J]. Pattern Recognition, 2024, 155: 110714.
[95]	徐彦威, 李军, 董元方, 等. YOLO系列目标检测算法综述[J]. 计算机科学与探索, 2024, 18(9): 2221-2238.
	XU Y W, LI J, DONG Y F, et al. Survey of development of YOLO object detection algorithms[J]. Journal of Frontiers of Computer Science & Technology, 2024, 18(9): 2221-2238.
[96]	KIALA Z, ODINDI J, MUTANGA O. Determining the capability of the tree-based pipeline optimization tool (TPOT) in mapping Parthenium weed using multi-date Sentinel-2 image data[J]. Remote Sensing, 2022, 14(7): 1687.
[97]	SU W H. Advanced machine learning in point spectroscopy, RGB- and hyperspectral-imaging for automatic discriminations of crops and weeds: A review[J]. Smart Cities, 2020, 3(3): 767-792.
[98]	HUANG G, XU Z B, LIU L, et al. Evaluating and deploying Large Vision-Language Models for fruit quality assessment in smart agriculture systems[J]. Computers and Electronics in Agriculture, 2025, 238: 110806.

主要步骤	详细操作
数据集准备	构建水果缺陷数据集，数据清洗，数据标注，数据集划分
模型设计	设计合适的深度学习模型，以适应水果缺陷检测任务的要求
模型训练	利用整理后的水果缺陷数据集训练模型，同时调整网络的训练参数
模型评估	选用合适的评估指标验证模型性能，如准确率、精确率、召回率、F ₁值、平均精度均值等
缺陷检测	将新的缺陷水果图像输入到训练好的模型中进行检测，并输出缺陷检测结果

	版本	贡献	框架
第一阶段	YOLOv1	单阶段检测器	Darknet
	YOLOv2	多尺度训练，维度聚类	Darknet
	YOLOv3	Darknet-53网络	Darknet
第二阶段	YOLOv4	空间金字塔池化模块，基于Mish的激活函数，CSPDarknet-53网络	Darknet
	YOLOv5	基于SWISH的激活函数，路径聚合网络	Pytorch
	YOLOv6	SimOTA标签分配策略，无锚框检测头	Pytorch
	YOLOv7	E-ELAN网络，Transformers预测头，扩展高效层聚合网络重参数化	Pytorch
	YOLOv8	分布焦点损失函数，空间金字塔快速池化层，任务对齐分配策略	Pytorch
第三阶段	YOLOv9	可编程梯度信息，广义高效层聚合网络	Pytorch
	YOLOv10	无非极大值抑制训练方法，双重标签分配，面向提升精度与效率的整体模型设计	Pytorch
	YOLOv11	带空间注意力的跨阶段局部网络，C3k2（Cross-Stage Partial Kernel-2）模块	Pytorch
	YOLOv12	残差高效层聚合网络，区域注意力机制	Pytorch
	YOLOv13	基于超图的自适应相关性增强，全流程聚合与分布	Pytorch
	YOLO26	MuSGD优化器，语义分割损失模块，多尺度原型模块	Pytorch

年份	模型	水果	改进方法	研究难点	mAP_0.5/%
2024	YOLOv7^［52］	柑橘	1）SIRI技术 2）坐标注意力机制 3）Mish激活函数 4）引入SPPFS（Spatial Pyramid Pooling - Fast， SPP Fand feature selection）模块	RGB图像无法检测内部缺陷	95.9
2022	YOLOv4^［53］	苹果	1）近红外相机图像采集系统 2）模型的通道剪枝与层剪枝	RGB图像无法检测内部缺陷	92.42
2023	YOLOv3^［54］	草莓	1）近红外相机图像采集系统 2）近红外与RGB图像像素融合	RGB图像无法检测内部缺陷	87.18
2023	YOLOv5^［55］	柑橘	1）双光源采集系统 2）广义交并比（Generalized Intersection over Union，GIoU）损失函数 3）卷积块注意力模块（Convolutional Block Attention Module，CBAM）	自然光下缺陷不易检测	98.7
2026	YOLOv11^［56］	柑橘	1）紫外荧光成像	自然光下缺陷不易检测	96.9
2025	YOLOv12^［57］	苹果	1）滤波成像与色彩空间重建结合	自然光下缺陷不易检测	98.9
2023	YOLOv5^［58］	草莓	1）引入分类模型 2）优化损失函数	复杂环境缺陷不易检测	85.4
2023	YOLOv5^［59］	草莓	1）引入Deeplabv3语义分割网络 2）添加分支结构 3）引入可变形卷积	复杂环境缺陷不易检测	62.8

YOLO系列模型在水果缺陷检测的研究进展及发展趋势

Research Progress and Development Trends of YOLO Family of Algorithms in Fruit Defect Detection

在线阅读

知网下载

本地下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 98

相关文章 15

编辑推荐

Metrics

本文评价

[1]	苏玉杰, 李玥, 魏霖静, 吴兵, 郭林海, 剡斌, 周慧, 高玉红, 康亮河, 刘欢, 苏舜昌. 基于轻量化改进YOLOv11n-Seg的胡麻倒伏区域识别方法[J]. 智慧农业(中英文), 2026, 8(2): 35-47.
[2]	吴婷婷, 郭俊睿, 陶秋洁, 陈世华, 郭善利. YOLOv8n-SSND：改进的无人机航拍藜麦穗目标检测轻量模型[J]. 智慧农业(中英文), 2026, 8(2): 59-71.
[3]	赵丽成, 卢鑫羽, 吴茜, 任妮, 周玲莉, 程雅雯, 胡安琦, 戚超. 基于改进YOLOv10和LAMP通道剪枝的串番茄成熟度检测算法研发[J]. 智慧农业(中英文), 2026, 8(2): 133-146.
[4]	修贤超, 费士祺, 黄文倩, 李楠, 苗中华. 基于轻量化Mamba-YOLO模型的梨表面缺陷检测方法[J]. 智慧农业(中英文), 2026, 8(2): 147-157.
[5]	刘浩然, 王宇, 赵学观, 吴华瑞, 付豪, 庞树杰, 翟长远. CD-YOLO：基于改进YOLOv11s的田间胡萝卜苗检测方法[J]. 智慧农业(中英文), 2026, 8(2): 158-174.
[6]	张文博, 江一珏, 宋巍, 贺琪, 张文博. CAGE-YOLO：基于遥感影像的水产养殖网箱高密度小目标检测模型[J]. 智慧农业(中英文), 2026, 8(2): 98-117.
[7]	巫晨旭, 左浩龙, 李刚. 面向云遮挡与复杂地块场景下的跨模态注意力多源遥感农作物分类[J]. 智慧农业(中英文), 2026, 8(2): 118-132.
[8]	李澳强, 戴航宇, 郭亚. 基于双目视觉和改进YOLOv11-pose的河蟹水下原位质量估测方法[J]. 智慧农业(中英文), 2026, 8(1): 192-202.
[9]	肖瑞宏, 谭立新, 王日凤, 宋敏, 胡程喜. 基于改进YOLOv11n的多尺度茶叶病害检测方法[J]. 智慧农业(中英文), 2026, 8(1): 62-71.
[10]	武张斌, 何宁, 吴延东, 郭新宇, 温维亮. 点云数据驱动的玉米叶片生物量估算方法[J]. 智慧农业(中英文), 2026, 8(1): 156-166.
[11]	高光甫, 王启磊, 宋丽雯, 冯海宽, 时雷, 杨浩, 刘杨, 岳继博. 基于CornYOLO的冠层内玉米果穗目标检测方法[J]. 智慧农业(中英文), 2026, 8(1): 167-177.
[12]	姚晓通, 曲绍业. 基于改进YOLOv12s的辣椒叶片病虫害轻量化检测方法[J]. 智慧农业(中英文), 2026, 8(1): 1-14.
[13]	赵峻, 聂志刚, 李广, 刘佳宇. 基于无人机低空近景图像的玉米螟虫害检测方法[J]. 智慧农业(中英文), 2025, 7(6): 111-123.
[14]	曹玉莹, 刘银川, 高新悦, 贾银江, 董守田. LightTassel-YOLO：一种基于无人机遥感的玉米雄穗实时检测方法[J]. 智慧农业(中英文), 2025, 7(6): 96-110.
[15]	张运, 张露敏, 许广涛, 郝佳慧. 基于双分支与多尺度注意力机制的稻虾田遥感提取方法[J]. 智慧农业(中英文), 2025, 7(6): 185-195.

YOLO系列模型在水果缺陷检测的研究进展及发展趋势

​​Research Progress and Development Trends of YOLO Family of Algorithms in Fruit Defect Detection

在线阅读

知网下载

本地下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 98

相关文章 15

编辑推荐

Metrics

本文评价

Research Progress and Development Trends of YOLO Family of Algorithms in Fruit Defect Detection