基于深度学习与多尺度特征融合的烤烟烟叶分级方法

doi:10.13733/j.jcam.issn.20955553.2022.01.023

摘要/Abstract

摘要： 为实现烤烟等级的快速准确识别，降低人工分级中主观因素对分级结果的影响，提高烟叶分级的准确性和一致性，提出一种基于烤烟RGB图像和深度学习的多尺度特征融合的烟叶图像等级分类方法，采用ResNet50提取烟叶图像特征，并引入基于注意力机制的SE模块（压缩激发模块），增强不同通道特征的重要程度；同时，采用FPN（特征金字塔网络）对提取的由浅及深不同层级的烟叶特征进行融合，以实现烟叶多尺度特征的表达。采集皖南地区6 068个烤烟的正面和背面图像用于建模和分析。结果表明，提出的烟叶分级方法的分级正确率比经典CNN（卷积神经网络）高出5.21%，分级模型在新批次7个等级烟叶上的分级正确率为80.14%，相邻等级的分级正确率为91.50%。因此，采用RGB图像结合深度学习技术可实现烤烟烟叶等级的良好识别，可为烤烟烟叶收购等级评价提供一种新方法。

关键词: 烟叶分级, 深度学习, 图像分类, 特征融合, 特征金字塔网络, SE模块

Abstract: In order to realize the rapid and accurate recognition of fluecured tobacco grade, reduce the influence of subjective factors on the grading results during manual grading, and improve the accuracy and consistency of tobacco grading, a tobacco image grading method based on multiscale feature fusion of RGB image and deep learning was proposed. The attention mechanism (SENet) was introduced to enhance the importance of different channel features. Meanwhile, FPN (Feature Pyramid Network) was used to fuse the extracted tobacco leaf features at different levels from shallow to deep to realize the expression of tobacco leaf multiscale features. The front and the back images of a total of 6 068 fluecured tobacco in southern Anhui were collected for modeling and analysis. The results indicated that the tobacco grading accuracy of the proposed method was 5.21% higher than that of the classical CNN (Convolutional Neural Network). The final grading accuracy of the model was 80.14% on the new batch of 7 grades of tobacco leaves, and the adjacent grading accuracy was 91.50%. Therefore, using RGB images combined with deep learning technology can accurately identify the grade of fluecured tobacco leaves, which provides a new method for the evaluation of fluecured tobacco grades in purchasing.

Key words: tobacco grading, deep learning, image classification, feature fusion, feature pyramid network, SE Net

中图分类号:

鲁梦瑶, 周强, 姜舒文, 王聪, 陈栋, 陈天恩, . 基于深度学习与多尺度特征融合的烤烟烟叶分级方法[J]. 中国农机化学报, 2022, 43(1): 158-166.

Lu Mengyao, Zhou Qiang, Jiang Shuwen, Wang Cong, Chen dong, Chen Tianen. . Fluecured tobacco leaf grading method based on deep learning and multiscale feature fusion[J]. Journal of Chinese Agricultural Mechanization, 2022, 43(1): 158-166.

参考文献

［１］　李小兰, 孙建生, 梁伟. 构建卷烟工业烟叶原料质量保障体系的思考［J］. 广东农业科学, 2008(11): 142-144.
Li Xiaolan, Sun Jiansheng, Liang Wei. A study on building quality assurance system for industrial tobacco raw materials ［J］. Guangdong Agricultural Sciences, 2008(11): 142-144.
［2］　潘周云, 包正元, 柳强, 等. 贵州山区烤烟专业化散烟分级队伍建设的现状与对策［J］. 贵州农业科学, 2017, 45(9): 166-170.
Pan Zhouyun, Bao Zhengyuan, Liu Qiang, et al. Current situation and countermeasures of specialization fluecured tobacco grading team construction in mountainous area, Guizhou ［J］. Guizhou Agricultural Sciences, 2017, 45(9): 166-170.
［3］　王树林, 王国英. 烟叶等级质量的提升策略与成效分析［J］. 安徽农业科学, 2017, 45(31): 223-225.
Wang Shulin, Wang Guoying. Strategies and effects of upgrading tobacco grades quality ［J］. Journal of Anhui Agricultural Sciences, 2017, 45(31): 223-225.
［4］　Maccormac J K M. Online image processing for tobacco grading in Zimbabwe ［J］. IEEE, 1993: 327-331.
［5］　张建平, 吴守一, 方如明, 等. 农产品质量的计算机辅助检验与分级(第Ⅱ报)烟叶自动分级模型的建立与训练［J］. 农业工程学报, 1997, 13(4): 184-188.
［6］　Zhang F, Zhang X. Classification and quality evaluation of tobacco leaves based on image processing and fuzzy comprehensive evaluation ［J］. Sensors, 2011, 11(3): 2369-2384.
［7］　李胜. 烤烟烟叶图像特征提取和质量分级研究［D］. 长沙: 中南大学, 2011.
［8］　袁奎. 烟叶特征的视觉信息表达与分级算法研究［D］. 贵阳: 贵州大学, 2015.
［9］　Lecun Y, Bottou L, Bengio Y, et al. Gradientbased learning applied to document recognition ［J］. Proceedings of the IEEE, 1998, 86(11): 2278-2324.
［10］　Thenmozhi K, Reddy U S. Crop pest classification based on deep convolutional neural network and transfer learning ［J］. Computers and Electronics in Agriculture, 2019, 164: 104906.
［11］　Zhou L B, Huang T, Guan C Y, et al. Realtime detection of cole diseases and insect pests in wireless sensor networks ［J］. Journal of Intelligent and Fuzzy Systems, 2019, 37(16): 1-12.
［12］　Huang Y P, Wang T H, Basanta H. Using fuzzy mask RCNN model to automatically identify tomato ripeness ［J］. IEEE Access, 2020, 8: 207672-207682.
［13］　Zhengjun Q, Jian C, Yiying Z, et al. Variety identification of single rice seed using hyperspectral imaging combined with convolutional neural network ［J］. Applied Sciences, 2018, 8(2): 212.
［14］　Inkyu S, Ge Z, D Feras, et al. Deep fruits: A fruit detection system using deep neural networks ［J］. Sensors, 2016, 16(8): 1222.
［15］　Xi R, Hou J, Lou W. Potato bud detection with improved faster RCNN ［J］. Transactions of the ASABE, 2020, 63(3): 557-569.
［16］　Hu G, Qian L, Liang D, et al. Selfadversarial training and attention for multitask wheat phenoltyping ［J］. Applied Engineering in Agriculture, 2019, 35(6): 1009-1014.
［17］　Dasari S K, Prasad V. A novel and proposed comprehensive methodology using deep convolutional neural networks for flue cured tobacco leaves classification ［J］. International Journal of Information Technology, 2019, 11(1): 107-117.
［18］　王士鑫. 基于卷积神经网络（CNN）的烤烟烟叶质量分级研究［D］. 昆明: 云南师范大学, 2020.
Wang Shixin. Quality classification of fluecured tobacco leaf based on convolutional neural network (CNN) ［D］. Kunming: Yunnan Normal University, 2020.
［19］　GB 2635—1992, 烤烟［S］.
［20］　Jie H, Li S, Gang S, et al. Squeezeandexcitation networks ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018: 7132-7141.
［21］　Lin T Y, Dollar P, Girshick R, et al. Feature pyramid networks for object detection ［C］. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017), 2017: 936-944.
［22］　李青山, 矫海楠, 王传义, 等. 烟叶正背面颜色参数与色素和主要化学成分的关系研究［J］. 江苏农业科学, 2016, 44(8): 332-336.
［23］　张永安, 马彩娟, 刘立博, 等. 不同产地烤烟叶片颜色正反面色差及其对感官评吸质量的影响［J］. 安徽农业科学, 2020, 48(13): 198-201.
Zhang Yongan, Ma Caijuan, Liu Libo, et al. Color deviation of both sides of fluecured tobacco leaves from different areas and its effect on sensory quality ［J］. Journal of Anhui Agricultural Sciences, 2020, 48(13): 198-201.
［24］　Jia D, Wei D, Socher R, et al. ImageNet: A largescale hierarchical image database ［C］. 2009 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2009), 2009: 248-255.
［25］　He K, Zhang X, Ren S, et al. Deep residual learning for image recognition ［C］. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016), 2016: 770-778.
［26］　Simonyan K, Zisserman A. Very deep convolutional networks for largescale image recognition ［J］. CoRR, 2014, 1409(15): 1556-1563.
［27］　Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the inception architecture for computer vision ［J］. IEEE, 2016: 2818-2826.

[1]	李东升, 胡文泽, 兰玉彬, 范明洪, 李翠云, 朱俊科. 深度学习在杂草识别领域的研究现状与展望[J]. 中国农机化学报, 2022, 43(9): 137-144.
[2]	王晓婷, 赵展, 王阳, 李林, . 基于改进Mask R-CNN的植物表型智能检测算法[J]. 中国农机化学报, 2022, 43(8): 151-157.
[3]	孙逸飞, 柳平增, 张艳, 宋长青, 张大雷, 马学文. 基于Sentinel-2A遥感影像的潍坊市冬小麦种植面积提取研究[J]. 中国农机化学报, 2022, 43(7): 98-105.
[4]	冯继克, 郑颖, 李平, 李艳翠, 张自阳, 郭晓娟. 基于特征选择的小麦籽粒品种识别研究[J]. 中国农机化学报, 2022, 43(7): 116-123.
[5]	张三林, 张立萍, 郑威强, 郭壮, 付子强. 基于YOLOv5的核桃品种识别与定位[J]. 中国农机化学报, 2022, 43(7): 167-172.
[6]	史方青, 王虎林, 黄华. 基于卷积神经网络的马铃薯芽眼检测识别研究[J]. 中国农机化学报, 2022, 43(6): 159-165.
[7]	曹金凤, 沈大港, 郭继鸿, 刘鹏, 李策, 兰添贺. 基于CNN和SVM分类优化的大蒜鳞芽朝向识别研究[J]. 中国农机化学报, 2022, 43(5): 134-139.
[8]	雷雪梅, 张光强, 姚旗, 刘伟渭, 邱帅. 基于卷积神经网络的农机图像自动识别研究[J]. 中国农机化学报, 2022, 43(5): 140-147.
[9]	陈国防, 陈兆英, 王玉亮, 王金星, 范国强, 李汉卿, . 基于数据增强深度学习的苹果花检测方法研究[J]. 中国农机化学报, 2022, 43(5): 148-155.
[10]	张燕斌, 杜健民, 王圆, 皮伟强, 高新超. 基于无人机高光谱遥感和3D-ResNet的荒漠草原地物分类[J]. 中国农机化学报, 2022, 43(4): 66-73.
[11]	何颖, 陈丁号, 彭琳. 基于改进YOLOv5模型的经济林木虫害目标检测算法研究[J]. 中国农机化学报, 2022, 43(4): 106-115.
[12]	项新建, 宋晓敏, 郑永平, 王海波, 方政洋. 基于MobileNet-YOLO的嵌入式人脸检测研究#br# #br# [J]. 中国农机化学报, 2022, 43(4): 124-130.
[13]	高文强, 肖志云, . 多尺度特征融合1D—CNN的马铃薯植株高光谱数据地物分类和缺素识别#br#[J]. 中国农机化学报, 2022, 43(3): 111-119.
[14]	董萍, 卫梦华, 时雷, 郭伟. 迁移学习在玉米叶片病害识别中的研究与应用[J]. 中国农机化学报, 2022, 43(3): 146-152.
[15]	张继成, 李德顺. 基于深度残差学习的成熟草莓识别方法[J]. 中国农机化学报, 2022, 43(2): 136-142.