[1]林朝剑,张广群,杨洁,等.基于迁移学习的林业业务图像识别[J].南京林业大学学报(自然科学版),2020,44(4):215-221.[doi:10.3969/j.issn.1000-2006.201904004]
 LIN Chaojian,ZHANG Guangqun,YANG Jie,et al.Transfer learning based recognition for forestry business images[J].Journal of Nanjing Forestry University(Natural Science Edition),2020,44(4):215-221.[doi:10.3969/j.issn.1000-2006.201904004]
点击复制

基于迁移学习的林业业务图像识别
分享到:

《南京林业大学学报(自然科学版)》[ISSN:1000-2006/CN:32-1161/S]

卷:
44
期数:
2020年4期
页码:
215-221
栏目:
研究论文
出版日期:
2020-09-01

文章信息/Info

Title:
Transfer learning based recognition for forestry business images
文章编号:
1000-2006(2020)04-0215-07
作者:
林朝剑 张广群 杨洁 徐鹏 李英杰 汪杭军
作者单位:浙江农林大学信息工程学院,浙江 杭州 311300; 浙江农林大学暨阳学院,浙江 诸暨 311800
Author(s):
LIN Chaojian1 ZHANG Guangqun1 YANG Jie2 XU Peng2 LI Yingjie2 WANG Hangjun2
(1.School of Information Engineering, Zhejiang A & F University, Hangzhou 311300, China; 2.Jiyang College,Zhejiang A & F University, Zhuji 311800, China)
关键词:
林业业务图像 迁移学习 森林管护 卷积神经网络 图像识别
Keywords:
forestry business image transfer learning forest manage and protect convolutional neural network image recognition
分类号:
S718
DOI:
10.3969/j.issn.1000-2006.201904004
文献标志码:
A
摘要:
目的 林业业务图像的识别分类有利于林业管理部门对相关事件作出合理的处置方案及指挥调度决策,从而充分发挥护林员的作用,提升森林管护的水平,达到保护森林资源和生态安全的目的。 方法 提出了一种针对林业业务图像基于迁移学习的卷积神经网络(convolutional neural networks)自动分类模型。在经过大规模辅助图像数据集ImageNet预训练的4种卷积神经网络模型的基础上,使用林业业务图像数据对训练好的模型进行迁移学习,采用新的全连接层取代原始的全连接层,其他层参数保持不变。 结果 在建立的4个类别林业业务图像数据集上,4个预训练卷积神经网络结构的迁移学习模型都具有较高的分类正确率。其中,基于Inception?v3的迁移学习模型识别精度最高,达到96.4%。 结论 利用基于Inception?v3的迁移学习模型进行林业业务图像分类是可行的。相比传统的特征提取识别方法以及其他预训练模型,Inception?v3模型具有很强的分类能力,可以在森林管护中发挥更广泛的应用。
Abstract:
Objective The recognition and classification of forestry business images are helpful for forestry administrative departments to make reasonable disposal plans, instruct others, and communicate decisions for relevant events. Therefore, full information can be provided to the forest protection personnel, which will improve the level of forest management and protection. Then, the protection of forest resources and ecological security can be achieved. Method We proposed an automatic classification model of convolutional neural networks based on the transfer learning for forestry business images. Four convolutional neural network models were pre-trained using a large-scale auxiliary image dataset, called ImageNet. Then, the forestry business image data, with a relatively small number of images, were used to transfer the trained model. The classification accuracy was compared and analyzed to select the model with the best classification effect. In this study, four convolutional neural network models were selected.The forestry business image dataset used for the model training and testing contained four classes, including animal death (AD), forest harvesting (FH), forest fire (FF), and forest pests (FP). The dataset included a total of 280 images, with 70 images in each class. The forestry business image dataset was divided into a training and test set, which accounted for 90% and 10%, respectively. The 10-fold cross-validation method was used to obtain the final test accuracy rate. A new fully connected layer was used to replace the original fully connected layer, while the other layers remained unchanged. The transfer learning method was divided into two steps. The Adam optimization algorithm was used to train the models. In addition, to compare the impacts of different data enhancement methods on the classification effect, three experimental scenarios were used in this study. Result The transfer learning models that we used with the four pre-trained convolutional neural network structures all had a higher classification accuracy compared with the traditional method. In particular, Inception-v3 had the highest test accuracy rate, reaching 96.4%, and the unit inference time was only 1.38 s. Inception-v1’s accuracy rate was slightly higher than Inception-v2’s. VGG-16 had the lowest accuracy rate, at only 92.1%, and the unit inference time was also the longest, at 2.42 s. Concerning the four categories of forestry business image datasets, the Inception-v3-based transfer learning model had the highest recognition accuracy. Compared with the VGG-16 model, the classification accuracy of Inception-v3 had a better recognition performance at a shorter recognition time. The experimental results concerning the different data enhancement methods showed that, compared with the test accuracy rate without data enhancement, the test accuracy rate after enhancing the data by 4 times was still 96.4% after 6 000 training steps. However, the test accuracy rate was approximately 96.8% between 1 000 and 3 000 steps, which was a slight improvement. Comparing the curves of the models with no-data enhancement and data enhanced 16 times, we found that the test accuracy rate of the enhanced dataset decreased, and as the number of training iterations increased the test accuracy rate became lower. In addition, the fluctuation of the test accuracy rate curve of the dataset enhanced 16 times was more pronounced compared with that of the other two cases. To analyze the “understanding” of the forestry business image dataset based on the Inception-v3 transfer learning model, a confusion matrix was constructed. It was found that FH was more likely to be confused with AD, while FP was more likely to be confused with FH. In contrast, FF was rarely confused with the other classes. There were obvious red flame features on the FF images, which were rare in the images of the other classes, so it was easy to distinguish between them. However, the other three classes did not have such distinguishing features, so there was confusion on some occasions. For the traditional classification methods, the extracted features were manually selected and had poor generalization characteristics. In this study, the traditional methods had a relatively poor classification accuracy concerning forestry business image classification. Conclusion Compared with the traditional feature extraction recognition methods, the convolutional neural network model had stronger classification capabilities. By modifying the network structure, the accuracy rate can be effectively improved. Therefore, it is feasible to use convolutional neural network models based on the transfer learning for forestry business image classification. With the improved recognition rate of convolution neural network models, forest rangers can play a greater role in forest protection.

参考文献/References:

1 LI W G.Analysis of forest management and protection as well as afforestation[J].Heilongjiang Sci,2017,8(24):70-71.
2 SUN W,MA Z B,CAO S S,et al.Analysis on the characteristics of forest resource data[J].J Northwest For Univ,2014,29(6):200-206,233.
3 ZHAO Y D,HUANG H,YAN X F,et al.Design of forest management and protection system based on iridium communication technology[J].Trans Chin Soc Agric Mach,2016,47(1):324-330.
4 ZHANG G.Measures and suggestions on the construction of the forest guard team in Dalian [J].Mod Agric Sci Technol,2018(1):144-145.
5 HU H,YANG X Q,WU D L.Forest manage and protect with location based service and system research and development[J].J Central South Univ For Technol,2017,37(9):12-19.DOI:10.14067/j.cnki.1673-923x.2017.09.003.
6 LIU Y P, GONG R, XIE C X, et al. Study on information extraction of poplar resources using Landsat remotely sensed datasets [J].Journal of Jiangsu Forestry Science & Technology,2017,44(2):28-33. DOI:10.3969/j.issn.1001-7380.2017.02.007.
7 LIU J Z,WANG X F,WANG T.Automatic identification of tree species based on deep learning[J].J Nanjing For Univ(Nat Sci Ed),2020,44(1):138-143.DOI:10.3969/j.issn.1000-2006.201809004.
8 LIU Z H,QI H N,ZHANG G Q,et al.Wood identification method based on microstructure images in cross?section[J].Sci Silvae Sin,2013,49(11):116-121.
9 ZHANG G Q,LI Y J,WANG H J.Classification of forestry images based on the BoW Model[J].J Zhejiang A&F Univ,2017,34(5):791-797.
10 ZHANG Q C,YANG L T,CHEN Z K,et al.A survey on deep learning for big data[J].Inf Fusion,2018,42:146-157.DOI:10.1016/j.inffus.2017.10.006.
11 GAO X, ZHAO Y F, XIONG Q, et al. Identification of tree species based on transfer learning[J]. Forest Engineering, 2019,35(5):68-75. DOI:10.16270/j.cnki.slgc.2019.05.012.
12 KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[J].Commun ACM,2017,60(6):84-90.DOI:10.1145/3065386.
13 HUANG J,JIANG Z G,ZHANG H P,et al.Ship object detection in remote sensing images using convolutional neural networks[J].J Beijing Univ Aeronaut Astronaut,2017,43(9):1841-1848.DOI:10.13700/j.bh.1001-5965.2016.0755.
14 ZHOU M,SHI Z W,DING H P.Aircraft classification in remote?sensing images using convolutional neural networks[J].J Image Graph,2017,22(5):702-708.
15 PAN S J,YANG Q.A survey on transfer learning[J].IEEE Trans Knowl Data Eng,2010,22(10):1345-1359.DOI:10.1109/tkde.2009.191.
16 WANG K L,YUAN H C.Aquatic animal image classification method based on transfer learning[J].J Comput Appl,2018,38(5):1304-1308,1326.
17 CHENG M, WU G Q, YUAN M T.Transfer learning for software defect prediction[J].Acta Electron Sin,2016,44(1):115-122.
18 LU H,FU X,HE Y N,et al.Cultivated land information extraction from high resolution UAV images based on transfer learning[J].Trans Chin Soc Agric Mach,2015,46(12):274-279,284.
19 LI S,WEI Z H,ZHANG B C,et al.Target recognition using the transfer learning?based deep convolutional neural networks for SAR images[J].J Univ Chin Acad Sci,2018,35(1):75-83.
20 ZHANG S W,ZHANG Q Q,LI P. Apple disease identification based on improved deep convolutional neural network[J].J For Eng, 2019,4(4):107-112 .DOI:10.13360/j.issn.2096-1359.2019.04.016.
21 DENG J,DONG W,SOCHER R,et al.ImageNet:a large?scale hierarchical image database[C]//2009 IEEE Conference on Computer Vision and Pattern Recognition,June 20-25,2009.Miami,FL.IEEE,2009:248-255.DOI:10.1109/cvpr. 2009.5206848.
22 SZEGEDY C,LIU W,JIA Y Q,et al.Going deeper with convolutions[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),June 7-12,2015.Boston,MA,USA.IEEE,2015:1-9.DOI:10.1109/cvpr.2015.7298594.
23 SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition ,June 27-30,2016.VegasLas,NV,USA.IEEE,2016: 2818- 2826.DOI:10.1109/cvpr.2016.308.
24 SIMONYAN K, ZISSERMAN A. Very deep convolutionalnetworks for large?scale image recognition[C]// International Conference on Learning Representations. 2014: 1-14.
25 IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]// International Conference on International Conference on Machine Learning. (2019-06-24).http://arxiv.org/help/license. 2015:448-456.
26 PATEL V.Kalman?based stochastic gradient method with stop condition and insensitivity to conditioning[J].SIAM J Optim,2016,26(4):2620-2648.DOI:10.1137/15m1048239

备注/Memo

备注/Memo:
收稿日期:2019-04-01
更新日期/Last Update: 2020-08-13