Construction and Screening of Intelligent Grading Model of Cigar Leaves Based on Deep Learning

doi:10.11924/j.issn.1000-6850.casb2025-0426

Abstract

Abstract:

This study aims to address the challenges in the cigar leaf grading process in China, where the lack of mature intelligent grading methods has led to a reliance on manual operations, resulting in inefficiency and inconsistent standards. The goal is to ensure the quality of cigar leaf products. The ‘FX-01’ variety, the main cultivar in Longyan, Fujian, was used as the research material, and a dataset of 8637 images covering nine commonly used acquisition grades was collected. Five state-of-the-art deep learning models (Swin, ViT, ResNet, Beit and ConvNext) were employed to develop intelligent grading models for upper, middle, and lower leaves, respectively. The results showed that all models met the requirements for daily response speed, with the ConvNext and ViT models achieving the best performance on the middle leaf test set, with an average accuracy of 93.3%. These findings demonstrate the feasibility of deep learning-based image technology in the intelligent grading of cigar wrapper leaves and provide technical support and theoretical guidance for further system improvement and mobile deployment, laying a foundation for the automation and standardization of cigar production.

Key words: cigar, grading, image recognition, deep learning, model construction, intelligent grading model, ConvNext, vision Transformer (ViT)

DU Chaofan, WANG Ruiqi, WU Tianyi, SHEN Cuiyu, SHEN Fulong, LAI Rijun, LIN Xiaolu, MA Xudong, XIE Xiaofang. Construction and Screening of Intelligent Grading Model of Cigar Leaves Based on Deep Learning[J]. Chinese Agricultural Science Bulletin, 2025, 41(34): 157-164.

Figures/Tables 11

References 26

[1]	李爱军, 秦艳青, 代惠娟, 等. 国产雪茄烟叶科学发展刍议[J]. 中国烟草学报, 2012, 18(1):112-114.
[2]	荣仕宾, 李晶晶, 赵园园, 等. 雪茄芯叶人工发酵温湿度控制对化学成分和香味品质的影响[J]. 中国烟草学报, 2021, 27(2):109-116.
[3]	高娅北, 钟秋, 王松峰, 等. 雪茄茄衣晾制过程中烟叶颜色和含水量变化及其相关分析[J]. 中国烟草科学, 2019, 40(2):57-63.
[4]	王佩, 贾世伟, 冯俊升, 等. 雪茄烟叶晾制过程中颜色变化探讨[J]. 现代农业科技, 2022(7):180-185.
[5]	赵晨婕, 周康, 姜宇, 等. 油分判定在雪茄烟叶分选分级中的影响研究[J]. 现代农业科技, 2023(10):194-197.
[6]	GUO J M, PRASETYO H, SU H S. Image indexing using the color and bit pattern feature fusion[J]. Journal of visual communication and image representation, 2013, 24(8):1360-1379. doi: 10.1016/j.jvcir.2013.09.005 URL
[7]	TAMURA H, MORI S, YAMAWAKI T. Textural features corresponding to visual perception[J]. Transactions on systems, man, and cybernetics, 1978, 8(6):460-473. doi: 10.1109/TSMC.1978.4309999 URL
[8]	BARTOLINI I, CIACCIA P, PATELLA M. WARP: Accurate retrieval of shapes using phase of Fourier descriptors and time warping distance[J]. IEEE Computer society, 2005, 27(1):142-147.
[9]	LOWE D G. Distinctive image features from scale-invariant keypoints[J]. International journal of computer vision, 2004, 60(2):91-110. doi: 10.1023/B:VISI.0000029664.99615.94
[10]	IQBAL K, ODETAYO M O, JAMES A. Content-based image retrieval approach for biometric security using colour, texture and shape features controlled by fuzzy heuristics[J]. Journal of computer and system sciences, 2012, 78(4):1258-1277. doi: 10.1016/j.jcss.2011.10.013 URL
[11]	SULTANA M, GAVRILOVA M. A content based feature combination method for face recognition[J]. Advances in intelligent systems and computing, 2013, 226:197-206.
[12]	LIU Z, MAO H, WU C Y, et al. A ConvNet for the 2020s[J]. arXiv e-prints,2022:11976-11986.
[13]	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[J]. IEEE,2016:770-778.
[14]	Bao H, Dong L, Piao S, et al. BEIT: BERT pre-training of image transformers[J]. arXiv preprint, 2021.
[15]	LIU Z, LIN Y, CAO Y, et al. Swin Transformer: Hierarchical vision transformer using shifted windows[J]. IEEE, 2021:10012-10022.
[16]	DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An image is worth 16×16 words: Transformers for image recognition at scale[J]. International conference on learning representations, 2021.
[17]	傅隆生, 冯亚利, ELKAMIL T, 等. 基于卷积神经网络的田间多簇猕猴桃图像识别方法[J]. 农业工程学报, 2018, 34(2):205-211.
[18]	THIPWIMON-CHOMPOOKHAM, GONWIRAT S, LATA S, et al. Plant leaf image recognition using multiple-grid based local descriptor and dimensionality reduction approach[A].// The 3rd International Conference on Information Science and Systems[C]. 2020.
[19]	汪睿琪, 张炳辉, 顾钢, 等. 基于YOLOv5的鲜烟叶成熟度识别模型研究[J]. 中国烟草学报, 2023, 29(2):46-55.
[20]	ZHAO J, DU C, LI Y. YOLO-Granada: A lightweight attentioned YOLO for pomegranate fruit detection[J]. Scientific reports, 2024, 14:16848. doi: 10.1038/s41598-024-67526-4
[21]	SHI Y, QING S, ZHAO L, et al. YOLO-Peach: A high-performance lightweight YOLOv8s-based model for accurate recognition and enumeration of peach seedling fruits[J]. Agronomy, 2024, 14(6):1628. doi: 10.3390/agronomy14081628 URL
[22]	WU M, LIN H, SHI X, et al. MTS-YOLO: A multi-task lightweight and efficient model for tomato fruit bunch maturity and stem detection[J]. Horticulturae, 2024, 10:1006. doi: 10.3390/horticulturae10091006 URL
[23]	DANVEER R, RANJAN A G. Ensemble of deep learning and machine learning approach for classification of handwritten Hindi numerals[J]. Journal of engineering and applied science, 2023, 70(1).
[24]	LI J, LI M. Flowering index intelligent detection of spray rose cut flowers using an improved YOLOv5s model[J]. Applied sciences, 2024, 14:9879. doi: 10.3390/app14219879 URL
[25]	LI Y, LI Y, LIANG C, et al. Improving deep learning with more data[J]. arXiv preprint, 2020.
[26]	申萍, 童丹, 陈哲. 基于偏斜叶色分布模式的新鲜烟叶成熟度判别[J]. 烟草科技, 2021, 54(8):26-35.

指标	权重
成熟度	0.06
油分	0.08
身份	0.08
完整度	0.3
均匀性	0.2
光泽	0.08
脉相	0.15
叶片结构	0.05

指标	权重
成熟度	0.06
油分	0.08
身份	0.08
完整度	0.3
均匀性	0.2
光泽	0.08
脉相	0.15
叶片结构	0.05

类别	条目	版本
硬件配置	系统	Ubuntu 18.04.1
硬件配置	显卡	GeForce RTX 3090
软件配置	Python	3.8
	深度学习框架	Pytorch
	CUDA	11.2

类别	条目	版本
硬件配置	系统	Ubuntu 18.04.1
硬件配置	显卡	GeForce RTX 3090
软件配置	Python	3.8
	深度学习框架	Pytorch
	CUDA	11.2

叶位	标签	训练集	验证集	测试集	总计
上部叶	JYB1	505	200	254	954
	JYB2	507	207	178	892
	JYB3	510	210	224	944
中部叶	JYC1	800	200	200	1200
	JYC2	600	284	207	1091
	JYC3	562	200	126	888
下部叶	JYX1	485	167	195	892
	JYX2	538	161	174	873
	JYX3	537	170	196	903