High-precision recognition method for imperfect grain images based on TransNeXt

doi:10.11772/j.issn.1001-9081.2025050593

Journal of Computer Applications ›› 2026, Vol. 46 ›› Issue (5): 1684-1691.DOI: 10.11772/j.issn.1001-9081.2025050593

• Frontier and comprehensive applications • Previous Articles

High-precision recognition method for imperfect grain images based on TransNeXt

Miaomiao YUAN¹, Yihong CHU², Guanjun YIN², Chunhua DENG¹^,²()

^1.School of Computer Science and Technology，Wuhan University of Science and Technology，Wuhan Hubei 430065，China
^2.Thermowell （Hubei） Intelligent Technology Company Limited，Wuhan Hubei 430223，China

Received:2025-05-29 Revised:2025-07-25 Accepted:2025-08-26 Online:2025-08-28 Published:2026-05-10
Contact: Chunhua DENG
About author:YUAN Miaomiao， born in 1999， M. S. candidate. Her research interests include computer vision， machine learning.
CHU Yihong， born in 1980， M. S.， engineer. His research interests include grain and oil inspection.
YIN Guanjun， born in 1986， M. S.， engineer. His research interests include intelligent control， machine vision.
Supported by:
Key Research and Development Program of Hubei Province(2023BAB071)

基于TransNeXt的粮食不完善粒图像高精度识别方法

袁淼淼¹, 褚毅宏², 尹冠军², 邓春华¹^,²()

^1.武汉科技大学计算机科学与技术学院，武汉 430065
^2.赛默威（湖北）智能科技有限公司，武汉 430223

通讯作者: 邓春华
作者简介:袁淼淼（1999—），女，湖北荆州人，硕士研究生，主要研究方向：计算机视觉、机器学习
褚毅宏（1980—），男，湖北潜江人，工程师，硕士，主要研究方向：粮油检测
尹冠军（1986—），男，湖北汉川人，工程师，硕士，主要研究方向：智能控制、机器视觉
基金资助:
湖北省重点研发计划项目(2023BAB071)

Abstract

Abstract:

Existing deep learning methods still face the following challenges in the research on high-precision imperfect grain recognition： the key discriminative features of imperfect grains are often distributed across image regions of varying scales and random positions， making it difficult to perceive these regions stably and comprehensively； meanwhile， the fine-grained discriminative features of multiple imperfect grains have diverse representations， and a unified modeling path struggles to optimize recognition performance of all categories simultaneously. To address these issues， a globally guided two-stage local feature learning framework was proposed based on TransNeXt. Deep representations of key discriminative regions were extracted under holistic perception and further refined through fine-grained modeling. Independently optimized network branches were designed for different categories， with all branches sharing the backbone to enable efficient adaptation and lightweight scalability. To support the above methods， an imperfect grain dataset covering multiple grain varieties with standardized category and discriminative region location annotations was constructed. Experimental results show that the proposed method achieves accuracy of 99.62% on the test set， verifying the framework's effectiveness and scalability in complex fine-grained image recognition tasks.

Key words: imperfect grain recognition, TransNeXt, fine-grained image recognition, local feature learning, multi-branch network

摘要：

现有深度学习方法在高精度粮食不完善粒识别研究中仍面临以下问题：不完善粒的关键判别特征局部分布于尺度不一、位置随机的图像区域，模型难以稳定、全面地感知这些区域；多类不完善粒的细粒度判别特征表达形式多样，统一建模路径难以同时优化所有类别的识别性能。因此，基于TransNeXt架构提出了一种全局引导的两阶段局部特征学习框架，在整图感知的基础上提取关键判别区域的深层表征，并在该区域上完成细粒度特征建模。进一步针对不同类别构建独立优化的网络分支，多个分支共享主干网络，实现对多类目标的高效适配与轻量扩展。为支撑上述方法，构建了一个覆盖多粮食品种并具备标准化类别与判别区域位置标注的不完善粒数据集。实验结果表明，所提方法在测试集上达到了99.62%的准确率，验证了它在复杂细粒度图像识别任务中的有效性与可扩展性。

关键词: 不完善粒识别, TransNeXt, 细粒度图像识别, 局部特征学习, 多分支网络

CLC Number:

TP391.41

Miaomiao YUAN, Yihong CHU, Guanjun YIN, Chunhua DENG. High-precision recognition method for imperfect grain images based on TransNeXt[J]. Journal of Computer Applications, 2026, 46(5): 1684-1691.

袁淼淼, 褚毅宏, 尹冠军, 邓春华. 基于TransNeXt的粮食不完善粒图像高精度识别方法[J]. 《计算机应用》唯一官方网站, 2026, 46(5): 1684-1691.

Figures/Tables 8

References 34

[1]	SHAO Y， GAO C， XUAN G， et al. Determination of damaged wheat kernels with hyperspectral imaging analysis［J］. International Journal of Agricultural and Biological Engineering， 2020， 13（5）： 194-198.
[2]	HAN K， ZHANG N， XIE H， et al. An improved strategy of wheat kernel recognition based on deep learning［J］. DYNA， 2023， 98（1）： 91-97.
[3]	陈卫东，范冰冰，王莹，等.基于机器视觉的谷物品种识别研究进展［J］.河南工业大学学报（自然科学版），2024，45（1）：133-142.
	CHEN W D， FAN B B， WANG Y， et al. Research progress of cereal variety identification based on machine vision［J］. Journal of Henan University of Technology （Natural Science Edition）， 2024， 45（1）： 133-142.
[4]	王昊，祝玉华，李智慧，等. 机器视觉在农作物种子检测中的研究进展［J］. 计算机工程与应用， 2023， 59（22）： 69-83.
	WANG H， ZHU Y H， LI Z H， et al. Research progress of machine vision in crop seed inspection［J］. Computer Engineering and Applications， 2023， 59（22）： 69-83.
[5]	LU J， TAN L， JIANG H. Review on Convolutional Neural Network （CNN） applied to plant leaf disease classification［J］. Agriculture， 2021， 11（8）： No.707.
[6]	STRUDEL R， GARCIA R， LAPTEV I， et al. Segmenter： Transformer for semantic segmentation［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 7242-7252.
[7]	DOSOVITSKIY A， BEYER L， KOLESNIKOV A， et al. An image is worth 16x16 words： Transformers for image recognition at scale［EB/OL］. ［2025-01-05］..
[8]	SHI D. TransNeXt： Robust foveal visual perception for vision transformers［C］// Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2024： 17773-17783.
[9]	ZHU Y， WANG H， LI Z， et al. Detection of corn unsound kernels based on GAN sample enhancement and improved lightweight network［J］. Journal of Food Process Engineering， 2024， 47（1）： No.e14499.
[10]	FU J， ZHENG H， MEI T. Look closer to see better： recurrent attention convolutional neural network for fine-grained image recognition［C］// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2017： 4476-4484.
[11]	KHAN A， VIBHUTE A D， MALI S， et al. A systematic review on hyperspectral imaging technology with a machine and deep learning methodology for agricultural applications［J］. Ecological Informatics， 2022， 69： No.101678.
[12]	罗建豪，吴建鑫.基于深度卷积特征的细粒度图像分类研究综述［J］.自动化学报，2017，43（8）：1306-1318.
	LUO J H， WU J X. A survey on fine-grained image categorization using deep convolutional features［J］. Acta Automatica Sinica， 2017， 43（8）： 1306-1318.
[13]	ZENG W， LI M. Crop leaf disease recognition based on self-attention convolutional neural network［J］. Computers and Electronics in Agriculture， 2020， 172： No.105341.
[14]	HU K， ZHANG H， LYU B， et al. Rapid detection of imperfect wheat grains based on deep learning technique［C］// Proceedings of the 2023 International Conference on Image Processing， Computer Vision and Machine Learning. Piscataway： IEEE， 2023： 821-828.
[15]	ZHOU Q， HUANG Z， ZHENG S， et al. A wheat spike detection method based on Transformer［J］. Frontiers in Plant Science， 2022， 13： No.1023924.
[16]	贺杰安，吴晓红，何小海，等.结合图像增强和卷积神经网络的小麦不完善粒识别［J］.计算机应用，2021，41（3）：911-916.
	HE J A， WU X H， HE X H， et al. Imperfect wheat kernel recognition combined with image enhancement and conventional neural network［J］. Journal of Computer Applications， 2021， 41（3）： 911-916.
[17]	WOO S， PARK J， LEE J Y， et al. CBAM： convolutional block attention module［C］// Proceedings of the 2018 European Conference on Computer Vision， LNCS 11211. Cham： Springer， 2018： 3-19.
[18]	ZHANG W， MA H， LI X， et al. Imperfect wheat grain recognition combined with an attention mechanism and residual network［J］. Applied Sciences， 2021， 11（11）： No.5139.
[19]	FU X， MA Q， YANG F， et al. Crop pest image recognition based on the improved ViT method［J］. Information Processing in Agriculture， 2024， 11（2）： 249-259.
[20]	CHEN J， LUO T， WU J， et al. A Vision Transformer network SeedViT for classification of maize seeds［J］. Journal of Food Process Engineering， 2022， 45（5）： No.e13998.
[21]	LIU Z， LIN Y， CAO Y， et al. Swin Transformer： hierarchical vision Transformer using shifted windows［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 9992-10002.
[22]	BI C， HU N， ZOU Y， et al. Development of deep learning methodology for maize seed variety recognition based on improved Swin Transformer［J］. Agronomy， 2022， 12（8）： No.1843.
[23]	ÇOLAK M， ÖZKAN Ö， ALMAN N P， et al. Classifying sunn pest damaged and healthy wheat grains across different species with YOLOV8 and Vision Transformers［J］. Journal of Advanced Research in Natural and Applied Sciences， 2024， 10（4）： 771-785.
[24]	KIM Y H， PARK K R. MTS-CNN： multi-task semantic segmentation-convolutional neural network for detecting crops and weeds［J］. Computers and Electronics in Agriculture， 2022， 199： No.107146.
[25]	SELVARAJU R R， COGSWELL M， DAS A， et al. Grad-CAM： visual explanations from deep networks via gradient-based localization［J］. International Journal of Computer Vision， 2020， 128（2）： 336-359.
[26]	ZHAO Y， LI J， CHEN X， et al. Part-guided relational Transformers for fine-grained visual recognition［J］. IEEE Transactions on Image Processing， 2021， 30： 9470-9481.
[27]	FAN L， DING Y， FAN D， et al. GrainSpace： a large-scale dataset for fine-grained and domain-adaptive recognition of cereal grains［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 21084-21093.
[28]	FAN L， DING Y， FAN D， et al. Identifying the defective： detecting damaged grains for cereal appearance inspection［C］// Proceedings of the 26th European Conference on Artificial Intelligence/ 12th Conference on Prestigious Applications of Intelligent Systems. Amsterdam： IOS Press， 2023： 660-667.
[29]	HE K， ZHANG X， REN S， et al. Deep residual learning for image recognition［C］// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2016： 770-778.
[30]	TAN M， LE Q V. EfficientNet： rethinking model scaling for convolutional neural networks［C］// Proceedings of the 36th International Conference on Machine Learning. New York： JMLR.org， 2019： 6105-6114.
[31]	RADOSAVOVIC I， KOSARAJU R P， GIRSHICK R， et al. Designing network design spaces［C］// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2020： 10425-10433.
[32]	LIU Z， MAO H， WU C Y， et al. A ConvNet for the 2020s［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 11966-11976.
[33]	TOUVRON H， CORD M， DOUZE M， et al. Training data-efficient image Transformers & distillation through attention［C］// Proceedings of the 38th International Conference on Machine Learning. New York： JMLR.org， 2021： 10347-10357.
[34]	HATAMIZADEH A， HEINRICH G， YIN H， et al. FasterViT： fast Vision Transformers with Hierarchical attention［EB/OL］. ［2025-02-25］. .

方法	准确率	精确率	召回率	F1分数
ResNet-50^［29］	99.03	99.06	99.03	99.04
EfficientNet-B3^［30］	99.23	99.24	99.23	99.23
RegNetX-3.2GF^［31］	99.27	99.28	99.27	99.27
ConvNeXt-T^［32］	99.45	99.46	99.45	99.45
DeiT-S^［33］	99.21	99.23	99.21	99.22
Swin-T^［21］	99.41	99.43	99.41	99.42
FasterViT-0^［34］	99.26	99.27	99.26	99.26
TransNeXt-Micro^［8］	99.48	99.48	99.48	99.48
本文方法	99.62	99.61	99.62	99.61

方法	准确率	精确率	召回率	F1分数
ResNet-50^［29］	99.03	99.06	99.03	99.04
EfficientNet-B3^［30］	99.23	99.24	99.23	99.23
RegNetX-3.2GF^［31］	99.27	99.28	99.27	99.27
ConvNeXt-T^［32］	99.45	99.46	99.45	99.45
DeiT-S^［33］	99.21	99.23	99.21	99.22
Swin-T^［21］	99.41	99.43	99.41	99.42
FasterViT-0^［34］	99.26	99.27	99.26	99.26
TransNeXt-Micro^［8］	99.48	99.48	99.48	99.48
本文方法	99.62	99.61	99.62	99.61

粮食品种	虫蚀粒				生芽粒				烧头粒
	Baseline		本文方法		Baseline		本文方法		Baseline		本文方法
	精确率	召回率	精确率	召回率	精确率	召回率	精确率	召回率	精确率	召回率	精确率	召回率
小麦	98.47	98.79	98.25	99.84	98.58	98.57	99.73	99.75	98.71	98.65	98.71	99.83
高粱	98.85	98.85	99.05	99.43	99.90	99.61	99.90	99.71	—	—	—	—
糙米	86.67	81.25	93.33	87.50	98.80	96.93	98.58	98.87	—	—	—	—
大麦	99.58	99.66	99.98	99.98	91.53	95.05	94.80	97.25	—	—	—	—
玉米	94.81	92.70	97.97	93.97	—	—	—	—	—	—	—	—
大米	95.24	86.96	95.45	91.30	—	—	—	—	—	—	—	—
大豆	98.05	99.21	99.80	99.61	—	—	—	—	—	—	—	—

粮食品种	虫蚀粒				生芽粒				烧头粒
	Baseline		本文方法		Baseline		本文方法		Baseline		本文方法
	精确率	召回率	精确率	召回率	精确率	召回率	精确率	召回率	精确率	召回率	精确率	召回率
小麦	98.47	98.79	98.25	99.84	98.58	98.57	99.73	99.75	98.71	98.65	98.71	99.83
高粱	98.85	98.85	99.05	99.43	99.90	99.61	99.90	99.71	—	—	—	—
糙米	86.67	81.25	93.33	87.50	98.80	96.93	98.58	98.87	—	—	—	—
大麦	99.58	99.66	99.98	99.98	91.53	95.05	94.80	97.25	—	—	—	—
玉米	94.81	92.70	97.97	93.97	—	—	—	—	—	—	—	—
大米	95.24	86.96	95.45	91.30	—	—	—	—	—	—	—	—
大豆	98.05	99.21	99.80	99.61	—	—	—	—	—	—	—	—

High-precision recognition method for imperfect grain images based on TransNeXt

基于TransNeXt的粮食不完善粒图像高精度识别方法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 8

References 34

Related Articles 4

Recommended Articles

Metrics

[1]	Shengwei XU, Jianbo WANG, Jijie HAN, Yijie BAI. Face forgery detection method based on tri-branch feature extraction [J]. Journal of Computer Applications, 2026, 46(4): 1292-1299.
[2]	Ailing QI, Xuanlin WANG. Fine-grained image recognition based on mid-level subtle feature extraction and multi-scale feature fusion [J]. Journal of Computer Applications, 2023, 43(8): 2556-2563.
[3]	Zhixiong ZHENG, Jianhua LIU, Shuihua SUN, Ge XU, Honghui LIN. Aspect-based sentiment analysis model fused with multi-window local information [J]. Journal of Computer Applications, 2023, 43(6): 1796-1802.
[4]	Taiming FU, Yan CHEN, Taoshen LI. Hard-negative sample mining for metric learning based on linear assignment [J]. Journal of Computer Applications, 2020, 40(2): 352-357.