吉林大学学报(工学版) ›› 2021, Vol. 51 ›› Issue (6): 2174-2181.doi: 10.13229/j.cnki.jdxbgxb20200749

• 计算机科学与技术 • 上一篇    

基于显著性检测的害虫图像分类

赵宏伟1(),霍东升1,王洁2,李晓宁3()   

  1. 1.吉林大学 计算机科学与技术学院，长春 130012
    2.上海航天控制技术研究所，上海 201109
    3.长春师范大学 计算机科学与技术学院，长春 130032
  • 收稿日期:2020-09-26 出版日期:2021-11-01 发布日期:2021-11-15
  • 通讯作者: 李晓宁 E-mail:zhaohw@jlu.edu.cn;lixiaoning@ccsfu.edu.cn
  • 作者简介:赵宏伟（1962-），男，教授，博士生导师. 研究方向：嵌入式人工智能. E-mail：zhaohw@jlu.edu.cn
  • 基金资助:
    吉林省省级科技创新专项资金项目(20190302026GX);吉林省自然科学基金项目(20200201037JC);中国高校科技期刊研究会青年基金项目(CUJS-QN-2021-049);吉林省教育厅职业教育与成人教育项目(2019ZCY403);吉林省教育厅科研项目(JJKH20181180KJ)

Image classification of insect pests based on saliency detection

Hong-wei ZHAO1(),Dong-sheng HUO1,Jie WANG2,Xiao-ning LI3()   

  1. 1.College of Computer Science and Technology，Jilin University，Changchun 130012，China
    2.Shanghai Academy of Spaceflight Technology，Shanghai 201109，China
    3.College of Computer Science and Technology，Changchun Normal University，Changchun 130032，China
  • Received:2020-09-26 Online:2021-11-01 Published:2021-11-15
  • Contact: Xiao-ning LI E-mail:zhaohw@jlu.edu.cn;lixiaoning@ccsfu.edu.cn

RICH HTML

 12   

PDF (PC)

174

摘要：

针对病虫害分类时害虫种类多样，且类间与类内差异大等问题，提出了一种害虫分类模型PestNet。模型主要由目标定位模块OPM和多特征融合模块MFFM组成，OPM通过U型网络结构整合害虫图像浅层细节信息和深层空间信息，初步划定显著区域并输出空间语义特征。MFFM通过对空间语义特征和抽象语义特征进行双线性池化操作，弱化背景信息，增加细节特征。此外，通过目标区域裁剪和掩膜等方式辅助训练模型，提高模型分类精度。将该模型在病虫害数据集IP102上进行实验，分类准确率可达77.40%，能够实现复杂背景下大规模害虫图像的分类识别。

关键词: 计算机应用, 病虫害分类, 显著性检测, 多特征融合, 数据增强

Abstract:

Due to the diverse species of pests and the large differences between classes, it is difficult to achieve accurate classification. Based on the two processes of target positioning and target recognition when the human visual system recognizes objects, a pest classification model PestNet is designed. The model is mainly composed of Object Positioning Module (OPM) and Multi-Feature Fusion Module (MFFM). OPM integrates the shallow detail information and deep spatial information of the image through a U-shaped network structure, preliminarily delineating significant areas, and outputting spatial semantic features. MFFM performs bilinear pooling operations on spatial semantic features and abstracts semantic features to weaken background information and increase detailed features. In addition, training is assisted by target area clipping and masking to improve model classification accuracy. We conducted experiments on the disease and insect pest data set IP102, and the model classification accuracy rate reached 77.40%, which can realize the classification and recognition of large-scale pest images under complex background.

Key words: computer application, insect pest classification, significance detection, multi-feature fusion, data enhancement

中图分类号: 

  • TP391

图1

PestNet模型架构"

图2

OPM模块架构"

图3

多模双线性池化"

图4

害虫数据集类间、类内差异"

图5

IP102数据集分布直方图"

图6

模型热力图"

表1

基础模型分类结果"

基础模型图像尺寸准确率/%
VGG-16［Mask］224×25667.90
Inception-V3［Mask］299×33170.88
ResNet-50［Mask］224×25671.57
ResNet-50［Mask］448×51273.27
ResNet-50［OPM］448×51274.34

表2

PestNet模型对照实验"

OPMMFFM裁剪掩膜准确率/%
√73.33
√74.34
√√75.61
√√√76.74
√√√76.03
√√√√77.40
1 Serre T, Wolf L, Bileschi S, et al. Robust object recognition with cortex-like mechanisms[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(3): 411-426.
2 Cheng X, Zhang Y H, Chen Y Q, et al. Pest identification via deep residual learning in complex background[J]. Computers and Electronics in Agriculture, 2017, 141: 351-356.
3 车颖, 冯皛, 郑宏亮. 基于卷积神经网络的超声造影图像去噪方法[J]. 吉林大学学报:理学版, 2021, 59(5): 1256-1259.
Che Ying, Feng Xiao, Zheng Hong-liang. Denoisng method of contrast-enhanced ultrasound image based on convolutional neural networks[J]. Journal of Jilin University Science Edition, 2021, 59(5): 1256-1259.
4 李伟伟. 非局部的可切换滤波图像椒盐噪声去除方法[J]. 吉林大学学报:理学版, 2019, 57(4): 910-916.
Li Wei-wei. Method for removing salt and pepper noise from non local switchable filter image[J]. Journal of Jilin University Science Edition, 2019, 57(4): 910-916.
5 Goodale M A, Milner A D. Separate visual pathways for perception and action[J]. Trends in Neurosciences, 1992, 15(1): 20-25.
6 郑雅羽, 田翔, 陈耀武. 基于时空特征融合的视觉注意模型[J]. 吉林大学学报: 工学版, 2009, 39(6): 1625-1630.
Zheng Ya-yu, Tian Xiang, Chen Yao-wu. Visual attention model based on fussion of spatiotemporal features[J]. Journal of Jilin University(Engineering and Technology Edition), 2009, 39(6): 1625-1630.
7 李鹏松, 李俊达, 吴良武, 等. 基于阈值分割法和卷积神经网络的图像识别算法[J]. 吉林大学学报:理学版, 2020, 58(6): 1436-1442.
Li Peng-song, Li Jun-da, Wu Liang-wu, et al. Image recognition algorithm based on threshold segmentation method and convolutional neural network[J]. Journal of Jilin University Science Edition, 2020, 58(6): 1436-1442.
8 傅博, 王瑞子, 王丽妍, 等. 基于深度卷积神经网络的水下偏色图像增强方法[J]. 吉林大学学报:理学版, 2021, 59(4): 891-899.
Fu Bo, Wang Rui-zi, Wang Li-yan, et al. Enhancement method of underwater color cast image based on deep convolutional neural network[J]. Journal of Jilin University Science Edition, 2021, 59(4): 891-899.
9 高云龙, 吴川, 朱明. 基于改进卷积神经网络的短文本分类模型[J]. 吉林大学学报:理学版, 2020, 58(4): 923-930.
Gao Yun-long, Wu Chuan, Zhu Ming. Short text classification model based on improved convolutional neural network[J]. Journal of Jilin University Science Edition, 2020, 58(4): 923-930.
10 Ronneberger O, Fischer P, Brox T. U-net: convolutional networks for biomedical image segmentation[C]∥International Conference on Medical Image Computing and Computer-Assisted Intervention, University of Freiburg, Germany, 2015: 234-241.
11 Goodfellow I J, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets[C]∥Proceedings of the 27th International Conference on Neural Information Processing Systems, Cambridge, MA, USA, 2014: 2672-2680.
12 Pinheiro P O, Collobert R. From image-level to pixel-level labeling with convolutional networks[C]∥2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, USA, 2015: 1713-1721.
13 Wang Z, Bovik A C, Sheikh H R, et al. Image quality assessment: from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4): 600-612.
14 范敏, 韩琪, 王芬, 等. 基于多层次特征表示的场景图像分类算法[J]. 吉林大学学报: 工学版, 2017, 47(6): 1909-1917.
Fan Min, Han Qi, Wang Fen, et al. Scene image categorization algorithm based on multi-level features representation[J]. Journal of Jilin University(Engineering and Technology Edition), 2017, 47(6): 1909-1917.
15 Wei X S, Wu J, Cui Q. Deep learning for fine-grained image analysis: a survey[J/OL]. [2020-09-18].
16 Wu X P, Zhan C, Lai Y K, et al. IP102: A large-scale benchmark dataset for insect pest recognition[C]∥2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, CA, USA, 2019: 8779-8788.
17 Selvaraju R R, Cogswell M, Das A, et al. Grad-cam: Visual explanations from deep networks via gradient-based localization[C]∥2017 IEEE International Conference on Computer Vision (ICCV), Venice, Italy, 2017: 618-626.
18 Hu T, Qi H, Huang Q, et al. See better before looking closer: Weakly supervised data augmentation network for fine-grained visual classification[J/OL]. [2020-09-20].
19 Liu C, Zoph B, Neumann M, et al. Progressive neural architecture search[C]∥Proceedings of the European Conference on Computer Vision, Munich, Germany, 2018: 1-16.
20 吴建,许镜,丁韬.基于集成迁移学习的细粒度图像分类算法[J].重庆邮电大学学报:自然科学版,2020,32(3):452-458.
Wu Jian, Xu Jing, Ding Tao. Fine-grained image classification algorithm based on ensemble methods of transfer learning[J]. JJournal of Chongqing University of Posts and Telecommunications(Natural Science Edition), 2020, 32(3): 452-458.
21 李伟, 黄鹤鸣, 武风英, 等. 基于深度多特征融合的自适应CNN图像分类算法[J]. 计算机工程, 2021, 47(9): 235-239, 251.
Li Wei, Huang He-ming, Wu Feng-ying, et al. Adaptive CNN-based image classification algorithm based on deep fusion of multi-feature[J]. Computer Engineering, 2021, 47(9): 235-239, 251.
22 生龙, 马建飞, 杨瑞欣, 等. 基于特征交换的CNN图像分类算法研究[J]. 计算机工程, 2020, 46(9): 268-273.
Sheng Long, Ma Jian-fei, Yang Rui-xin, et al. Research on CNN image classification algorithm based on feature exchange[J]. Computer Engineering, 2020, 46(9): 268-273.
[1] 王生生,陈境宇,卢奕南. 基于联邦学习和区块链的新冠肺炎胸部CT图像分割[J]. 吉林大学学报(工学版), 2021, 51(6): 2164-2173.
[2] 林俊聪,雷钧,陈萌,郭诗辉,高星,廖明宏. 基于电影视觉特性的动态多目标实时相机规划[J]. 吉林大学学报(工学版), 2021, 51(6): 2154-2163.
[3] 姚引娣,贺军瑾,李杨莉,谢荡远,李英. 自构建改进型鲸鱼优化BP神经网络的ET0模拟计算[J]. 吉林大学学报(工学版), 2021, 51(5): 1798-1807.
[4] 赵宏伟,张子健,李蛟,张媛,胡黄水,臧雪柏. 基于查询树的双向分段防碰撞算法[J]. 吉林大学学报(工学版), 2021, 51(5): 1830-1837.
[5] 曹洁,屈雪,李晓旭. 基于滑动特征向量的小样本图像分类方法[J]. 吉林大学学报(工学版), 2021, 51(5): 1785-1791.
[6] 孙小雪,钟辉,陈海鹏. 基于决策树分类技术的学生考试成绩统计分析系统[J]. 吉林大学学报(工学版), 2021, 51(5): 1866-1872.
[7] 张萌谡,刘春天,李希今,黄永平. 基于K⁃means聚类算法的绩效考核模糊综合评价系统设计[J]. 吉林大学学报(工学版), 2021, 51(5): 1851-1856.
[8] 王春波,底晓强. 基于标签分类的云数据完整性验证审计方案[J]. 吉林大学学报(工学版), 2021, 51(4): 1364-1369.
[9] 欧阳丹彤,刘扬,刘杰. 故障响应指导下基于测试集的故障诊断方法[J]. 吉林大学学报(工学版), 2021, 51(3): 1017-1025.
[10] 钱榕,张茹,张克君,金鑫,葛诗靓,江晟. 融合全局和局部特征的胶囊图神经网络[J]. 吉林大学学报(工学版), 2021, 51(3): 1048-1054.
[11] 朱小龙,谢忠. 基于机器学习的地理空间数据抽取算法[J]. 吉林大学学报(工学版), 2021, 51(3): 1011-1016.
[12] 孙宝凤,任欣欣,郑再思,李国一. 考虑工人负荷的多目标流水车间优化调度[J]. 吉林大学学报(工学版), 2021, 51(3): 900-909.
[13] 刘元宁,吴迪,朱晓冬,张齐贤,李双双,郭书君,王超. 基于YOLOv3改进的用户界面组件检测算法[J]. 吉林大学学报(工学版), 2021, 51(3): 1026-1033.
[14] 陈广秋,陈昱存,李佳悦,刘广文. 基于DNST和卷积稀疏表示的红外与可见光图像融合[J]. 吉林大学学报(工学版), 2021, 51(3): 996-1010.
[15] 王淑敏,陈伟. 基于连续密度隐马尔可夫模型的矿下异常行为识别算法[J]. 吉林大学学报(工学版), 2021, 51(3): 1067-1072.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
版权所有 © 《吉林大学学报(工学版)》编辑部
地址：长春市人民大街5988号　电话：+86-431-85095297　传真：+86-431-85094074　E-mail: xbgxb@jlu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发