Effective distributed convolutional neural network architecture for remote sensing images target classification with a pre-training approach

doi:10.21629/JSEE.2019.02.02

Abstract

Abstract:

How to recognize targets with similar appearances from remote sensing images (RSIs) effectively and efficiently has become a big challenge. Recently, convolutional neural network (CNN) is preferred in the target classification due to the powerful feature representation ability and better performance. However, the training and testing of CNN mainly rely on single machine. Single machine has its natural limitation and bottleneck in processing RSIs due to limited hardware resources and huge time consuming. Besides, overfitting is a challenge for the CNN model due to the unbalance between RSIs data and the model structure. When a model is complex or the training data is relatively small, overfitting occurs and leads to a poor predictive performance. To address these problems, a distributed CNN architecture for RSIs target classification is proposed, which dramatically increases the training speed of CNN and system scalability. It improves the storage ability and processing efficiency of RSIs. Furthermore, Bayesian regularization approach is utilized in order to initialize the weights of the CNN extractor, which increases the robustness and flexibility of the CNN model. It helps prevent the overfitting and avoid the local optima caused by limited RSI training images or the inappropriate CNN structure. In addition, considering the efficiency of the Naïve Bayes classifier, a distributed Naïve Bayes classifier is designed to reduce the training cost. Compared with other algorithms, the proposed system and method perform the best and increase the recognition accuracy. The results show that the distributed system framework and the proposed algorithms are suitable for RSIs target classification tasks.

Key words: convolutional neural network (CNN), distributed architecture, remote sensing images (RSIs), target classification, pre-training

Binquan LI, Xiaohui HU. Effective distributed convolutional neural network architecture for remote sensing images target classification with a pre-training approach[J]. Journal of Systems Engineering and Electronics, 2019, 30(2): 238-244.

Figures/Tables 14

Fig 1

Table 1

Fig 2

Fig 3

Fig 4

Table 2

Table 3

Table 4

Table 5

Fig 5

Table 6

Fig 6

Fig 7

Fig 8

References 25

1	LECUN Y, KAVUKCUOGLU K, FARABET C. Convolutional networks and applications in vision. Proc. of the IEEE International Symposium on Circuits & Systems, 2010, 253- 256.
2	SERMANET P, EIGEN D, ZHANG X, et al. OverFeat:integrated recognition, localization and detection using convolutional networks. Proc. of the International Conference on Learning Representations, 2014, 1- 16.
3	ZEILER M D, FERGUS R. Visualizing and understanding convolutional networks. Proc. of the European Conference on Computer Vision, 2014, 818- 833.
4	ZHANG F, DU B, ZHANG L. Saliency-guided unsupervised feature learning for scene classification. IEEE Trans. on Geoscience & Remote Sensing, 2015, 53 (4): 2175- 2184.
5	HU F, XIA G, WANG Z, et al. Unsupervised feature learning via spectral clustering of multidimensional patches for remotely sensed scene classification. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2015, 8 (5): 2015- 2030.
6	ZHAO L J, TANG P, HUO L Z. Land-use scene classification using a concentric circle-structured multi-scale bag-of-visualwords model. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2014, 7 (12): 4620- 4631.
7	CHEN S, TIAN Y. Pyramid of spatial relations for scene-level land use classification. IEEE Trans. on Geoscience & Remote Sensing, 2015, 53 (4): 1947- 1957.
8	PENATTI O A, NOGUEIRA K, SANTOS J A D. Do deep features generalize from everyday objects to remote sensing and aerial scenes domains? Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2015, 44- 51.
9	MARMANIS D, DATCU M, ESCH T, et al. Deep learning earth observation classification using image net pre-trained networks. IEEE Geoscience & Remote Sensing Letters, 2016, 13 (1): 105- 109.
10	SCOTT G J, ENGLAND M R, STARMS W A, et al. Training deep convolutional neural networks for land-cover classification of high-resolution imagery. IEEE Geoscience & Remote Sensing Letters, 2017, 14 (4): 549- 553.
11	BAO H C, FANG L, LIU R Y. Research and application of the storage way of land use change records. Journal of Zhejiang University (Science Edition), 2011, 38 (2): 218- 222.
12	LAI J B, LUO X L, YU T, et al. Remote sensing data organization model based on cloud computing. Computer Science, 2013, 40 (7): 80- 84.
13	YANG H P, SHEN Z F, LUO J C, et al. Recent developments in high performance geocomputation for massive remote sensing data. Journal of Geo-Information Science, 2013, 15 (1): 128- 136. doi: 10.3724/SP.J.1047.2013.000128
14	LIU Y, GUO W, JIANG W S, et al. Research of remote sensing service based on cloud computing mode. Application Research of Computers, 2009, 26 (9): 3428- 3431.
15	REN F H, WANG J N. Turning remote sensing to cloud services:technical research and experiment. Journal of Remote Sensing, 2012, 16 (6): 1339- 1346.
16	WAN B, YANG L. Data center:GIS function warehouse. Earth Science-Journal of China University of Geosciences, 2010, 35 (3): 357- 361. doi: 10.3799/dqkx.2010.039
17	CHU C, KIM S K, LIN Y A, et al. Map-reduce for machine learning on multicore. Proc. of Advances in Neural Information Processing Systems, 2007, 281- 288.
18	MACKAY D J C. A practical Bayesian framework for back propagation networks. Neural Computation, 1992, 4 (3): 448- 472. doi: 10.1162/neco.1992.4.3.448
19	SNOEK J, LAROCHELLE H, ADAMS R P. Practical Bayesian optimization of machine learning algorithms. Proc. of Advances in Neural Information Processing Systems, 2012, 2951- 2959.
20	RUMELHART D, HINTON G, WILLIAMS R. Learning representations by back-propagating errors. Nature, 1986, 323 (6088): 533- 536. doi: 10.1038/323533a0
21	XU M, ZENG G, XU X, et al. Application of Bayesian regularized BP neural network model for trend analysis, acidity and chemical composition of precipitation in north Carolina. Water, Air, and Soil Pollution, 2006, 172 (1-4): 167- 184. doi: 10.1007/s11270-005-9068-8
22	FORESEE F D, HAGAN M T. Gauss-Newton approximation to Bayesian regularization. Proc. of the International Joint Conference on Neural Networks, 1997, 1930- 1935.
23	ERHAN D. Why does unsupervised pre-training help deep learning?. Journal of Machine Learning Research, 2010, 11 (3): 625- 660.
24	FAN J, XU W, WU Y, et al. Human tracking using convolutional neural networks. IEEE Trans. on Neural Network, 2010, 21 (10): 1610- 1623. doi: 10.1109/TNN.2010.2066286
25	LIU B, BLASCH E, CHEN Y, et al. Scalable sentiment classification for big data analysis using Naïve Bayes classifier. Proc. of the IEEE International Conference on Big Data, 2013, 99- 104.

Procedure	Data
Procedure	Input	Output
Map stage	$\langle$Target Class ID, training sample$\rangle$	$\langle$weight $w, $ local $\Delta w\rangle$
Reduce stage	$\langle$weight $w, $ local $\Delta w\rangle$	$\langle$weight $w, $ global $\Delta w\rangle$
Main	$\langle$training sample, parameters of network$\rangle$	$\langle$weight file$\rangle$

Target	Training sample	Testing sample
Helicopter-1	10	3
Helicopter-2	15	4
Early warning aircraft	20	5
Bomber-1	29	8
Bomber-2	21	5
Bomber-3	4	1
Fighter-1	12	3
Fighter-2	19	5
Fighter-3	5	1
Transport plane-1	40	10
Transport plane-2	36	9
Airliner	284	71
Warship	60	15
Aircraft carrier	8	2
Freighter	214	53
Total	777$\times$8 = 6 216	195$\times$8 = 1 560

Performance	Method
Performance	CNN+Pre-training	CNN	SVM+SIFT
Average value of accuracy/%	93	86	76

Training method	Time/h
CNN	12.9
Distributed CNN	4.2

[1]	Zhengliang ZHU, Degui YANG, Junchao ZHANG, Feng TONG. Dataset of human motion status using IR-UWB through-wall radar [J]. Journal of Systems Engineering and Electronics, 2021, 32(5): 1083-1096.
[2]	Wantian WANG, Ziyue TANG, Yichang CHEN, Yongjian SUN. Parity recognition of blade number and manoeuvre intention classification algorithm of rotor target based on micro-Doppler features using CNN [J]. Journal of Systems Engineering and Electronics, 2020, 31(5): 884-889.
[3]	Baojun ZHAO, Boya ZHAO, Linbo TANG, Wenzheng WANG, Chen WU. Multi-scale object detection by top-down and bottom-up feature pyramid network [J]. Journal of Systems Engineering and Electronics, 2019, 30(1): 1-12.
[4]	Liangkui LIN, Shaoyou WANG, Zhongxing TANG. Using deep learning to detect small targets in infrared oversampling images [J]. Journal of Systems Engineering and Electronics, 2018, 29(5): 947-952.
[5]	Bendong Zhao, Huanzhang Lu, Shangfeng Chen, Junliang Liu, and Dongya Wu. Convolutional neural networks for time series classification [J]. Systems Engineering and Electronics, 2017, 28(1): 162-.
[6]	Xufeng Zhu, Caiwen Ma, Bo Liu, and Xiaoqian Cao. Target classification using SIFT sequence scale invariants [J]. Journal of Systems Engineering and Electronics, 2012, 23(5): 633-639.
[7]	Wang Yang, Lu Jiaguo & Wu Xianliang. New algorithm of target classification in polarimetric SAR [J]. Journal of Systems Engineering and Electronics, 2008, 19(2): 273-279.