基于CNN网络和多任务损失函数的实时叶片识别

doi:10.16182/j.issn1004731x.joss.19-VR0473

摘要/Abstract

摘要： 针对传统叶片识别易受环境干扰,难以实现复杂背景下的多叶片实时识别问题,提出一种基于CNN网络和多任务损失函数的实时叶片识别方法。采用CNN网络提取叶片图像特征图,输入到RPN网络生成区域候选框;依据特征图和区域候选框,提取候选框特征图,分别进行叶片分类和边界框回归,预测叶片类别和叶片预测框的定位;利用多任务损失函数约束分类和回归,来提高叶片分类和回归的准确率和运算速度。实验结果表明,该方法的平均实时叶片识别准确率为91.8%,平均实时识别速度为25 fps。

关键词: 叶片识别, 特征图, CNN网络, 多任务损失函数, 区域候选框

Abstract: Aiming at the problems that the traditional leaf recognition is susceptible to the environmental interference and is difficult to realize the multi-leaf real-time recognition in complex background, a real-time leaf recognition method based on CNN network and multi-task loss function is proposed. The CNN network is used to the extract image feature maps and input them into RPN network to generate regional proposals. On the basis of the feature maps and region proposals, the feature map is proposaled, the leaf classification and bounding box regression are performed respectively, and the leaf classification and position of the leaf prediction box are predicted. The multi-task loss function is used to constrain the classification and regression to improve the accuracy and speed of the leaf image classification and regression. Experimental results show that the average real-time leaf recognition accuracy is 91.8%, and the average real-time leaf recognition speed is 25 fps.

Key words: leaf recognition, feature map, CNN network, multitask loss function, region proposal

中图分类号:

TP391.9

蔡兴泉,涂宇欣,葛亚坤等 . 基于CNN网络和多任务损失函数的实时叶片识别[J]. 系统仿真学报, 2020, 32(7): 1279-1286.

Cai Xingquan,Tu Yuxin,Ge Yakun,et al . Real-time Leaf Recognition Method Based on CNN Network and Multi-task Loss Function[J]. Journal of System Simulation, 2020, 32(7): 1279-1286.

参考文献 24

[1]	张宁, 刘文萍. 基于图像分析的植物叶片识别技术综述[J]. 计算机应用研究, 2011, 28(11): 4001-4007.Zhang Ning, Liu Wenping.Plant Leaf Recognition Technology Based on Image Analysis[J]. Application Research of Computers, 2011, 28(11): 4001-4007.
[2]	Hu J, Chen Z B, Yang M, et al.A Multi-Scale Fusion Convolutional Neural Network for Plant Leaf Recognition[J]. IEEE Signal Processing Letters (S1070-9908), 2018, 25(6): 853-857.
[3]	杨天天, 潘晓星, 穆立蔷. 基于叶片图像特征数字化信息识别7种柳属植物[J]. 东北林业大学学报, 2014, 42(12): 75-79.Yang Tiantian, Pan Xiaoxing, Mu Liqiang.Identification of Seven Salix Species Using Digital Information Analysis of Leaf Image Characteristics[J]. Journal of Northeast Forestry University, 2014, 42(12): 75-79.
[4]	刘骥, 曹凤莲, 甘林昊. 基于叶片形状特征的植物识别方法[J]. 计算机应用, 2016, 36(S2): 200-202, 206.Liu Ji, Cao Fenglian, Gan Linhao.Plant Identification Method Based on Leaf Shape Features[J]. Journal of Computer Applications, 2016, 36(S2): 200-202, 206.
[5]	邹秋霞, 郜鲁涛, 盛立冲. 基于Android手机和图像特征识别技术的植物叶片分类系统的研究[J]. 安徽农业科学, 2015, 43(11): 367-369.Zou Qiuxia, Gao Lutao, Sheng Lichong.Study on Plant Leaves Classification System Based on Android Mobile Phone and Image Feature Recognition Technology[J]. Journal of Anhui Agricultural Sciences, 2015, 43(11): 367-369.
[6]	Thanikkal J G, Dubey A K, Thomas M T.Whether color, shape and texture of leaves are the key features for image processing based plant recognition? An analysis![C]// 2017 Recent Developments in Control, Automation & Power Engineering(RDCAPE). Piscataway, NJ: IEEE, 2017: 404-409.
[7]	Munisami T, Ramsurn M, Kishnah S, et al.Plant Leaf Recognition Using Shape Features and Colour Histogram with K-nearest Neighbour Classifiers[J]. Procedia Computer Science (S1877-0509), 2015, 58: 740-747.
[8]	Srivastava V, Khunteta A.Comparative Analysis of Leaf Classification and Recognition by different SVM Classifiers[C]// 2018 International Conference on Inventive Research in Computing Applications(ICIRCA). Piscataway, NJ: IEEE, 2018: 626-631.
[9]	叶继华, 时淑霞, 李汉曦. 基于深度学习的驾驶关注区域检测方法研究[J]. 系统仿真学报, 2019, 31(7): 1421-1428.Ye Jihua, Shi Shuxia, Li Hanxi.Research and Implementation of Driving Concern Area Detection Based on Deep Learning[J]. Journal of System Simulation, 2019, 31(7): 1421-1428.
[10]	Reyes A K, Caicedo J C, Camargo J E.Fine-tuning Deep Convolutional Networks for Plant Recognition[J]. CLEF(Working Notes), 2015, 1391: 467-475.
[11]	杜兰, 刘斌, 王燕. 基于卷积神经网络的SAR图像目标检测算法[J]. 电子与信息学报, 2016, 38(12): 3018-3025.Du Lan, Liu Bin, Wang Yan.Target Detection Method Based on Convolutional Neural Network for SAR Image[J]. Journal of Electronics & Information Technology, 2016, 38(12): 3018-3025.
[12]	吴昀璞, 金炜东, 黄颖坤. 基于多域融合 CNN 的高速列车转向架故障检测[J]. 系统仿真学报, 2018, 30(11): 4492-4497.Wu Yunpu, Jin Weidong, Huang Yingkun.Fault Diagnosis of High Speed Train Bogie Based on Multi-domain Fusion CNN[J]. Journal of System Simulation, 2018, 30(11): 4492-4497.
[13]	周飞燕, 金林鹏, 董军. 卷积神经网络研究综述[J]. 计算机学报, 2017, 40(6): 1229-1251.Zhou Feiyan, Jin Linpeng, Dong Jun.Review of Convolutional Neural Network[J]. Chinese Journal of Computers, 2017, 40(6): 1229-1251.
[14]	刘晶晶. 基于深度网络特征学习的植物叶片识别算法研究与实现[D]. 深圳: 深圳大学硕士论文, 2017.Liu Jingjing.Research and Implementation of Plant Leaf Recognition Algorithm Based on Deep Network Feature Learning[D]. Shenzhen: Shenzhen University, Master thesis, 2017.
[15]	张帅, 淮永建. 基于分层卷积深度学习系统的植物叶片识别研究[J]. 北京林业大学学报, 2016, 38(9): 108-115.Zhang Shuai, Huai Yongjian.Leaf Image Recognition Based on Layered Convolutions Neural Network Deep Learning[J]. Journal of Beijing Forestry University, 2016, 38(9): 108-115.
[16]	Uijlings J R R, van de Sande K E A, Gevers T. Selective Search for Object Recognition[J]. International Journal of Computer Vision (S1573-1405), 2013, 104(2): 154-171.
[17]	Chen M Y, Tang Y C, Zou X J.High-accuracy multi-camera reconstruction enhanced by adaptive point cloud correction algorithm[J]. Optics and Lasers in Engineering. Optics and Lasers in Engineering (S0143-8166), 2019, 122: 170-183.
[18]	Lin G C, Tang Y C, Zou X J.In-field citrus detection and localisation based on RGB-D image analysis[J]. Biosystems Engineering (S1537-5110), 2019, 186: 34-44.
[19]	Girshick R, Donahue J, Darrelland T, et al.Rich Feature Hierarchies for Object Detection and Semantic Segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ: IEEE, 2014.
[20]	曹诗雨, 刘跃虎, 李辛昭. 基于Fast R-CNN的车辆目标检测[J]. 中国图象图形学报, 2017, 22(5): 671-677.Cao Shiyu, Liu Yuehu, Li Xinzhao.Vehicle detection method based on fast R-CNN[J]. Journal of Image and Graphics, 2017, 22(5): 671-677.
[21]	Girshick R.Fast R-CNN[C]. Proceedings of the IEEE international conference on computer vision. Piscataway, NJ: IEEE, 2015: 1440-1448.
[22]	Redmon J, Divvala S, Girshick R, et al.You Only Look Once: Unified, Real-Time Object Detection[C]. Proceedings of the IEEE conference on computer vision and pattern recognition. Piscataway, NJ: IEEE, 2016: 779-788.
[23]	王全东, 常天庆, 张雷. 面向多尺度坦克装甲车辆目标检测的改进Faster R-CNN算法[J]. 计算机辅助设计与图形学学报, 2018, 30(12): 2278-2291.Wang Quandong, Chang Tianqing, Zhang Lei.An Improved Faster R-CNN Algorithm for Detection of Multi-scale Tank Armored Vehicle Targets[J]. Journal of Computer-Aided Design & Computer Graphics, 2018, 30(12): 2278-2291.
[24]	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 (S0162-8828), 2015, 39(6): 1137-1149.