Please wait a minute...
Remote Sensing for Land & Resources    2020, Vol. 32 Issue (3) : 80-89     DOI: 10.6046/gtzyyg.2020.03.11
Remote sensing image scene classification based on Inception-V3
CAI Zhiling1(), WENG Qian1(), YE Shaozhen1, JIAN Cairen2
1. College of Mathematics and Computer Science, Fuzhou University, Fuzhou 350116, China
2. School of Information Science and Technology, Xiamen University Tan Kahkee College, Zhangzhou 363105, China
Download: PDF(5410 KB)   HTML
Export: BibTeX | EndNote | Reference Manager | ProCite | RefWorks    

With the deepening and cross-fusion of modern remote sensing image research, the classification of high spatial resolution remote sensing image (referred to as “high-resolution image”) has become a research hotspot in the field of remote sensing. As the phenomenon of “homology spectrum” and “homology spectrum” of high-resolution image is more serious, the deep learning method that has emerged in recent years has proposed a new solution for high-resolution image classification. However, the lack of training samples of remote sensing images can easily lead to over-fitting of deep learning networks. In this paper, an improved Inception-V3 remote sensing image scene classification model is proposed by using deep learning method and transfer learning strategy. The model first adds Dropout layer before the full connection layer of the original Inception-V3 model in order to avoid over-fitting. In the training process, the transfer learning strategy is adopted to make full use of the existing model and knowledge and improve the training efficiency. The experimental results based on AID and NWPU-RESISC45 datasets show that the improved Inception-V3 has faster convergence speed and smoother training effect than the original Inception-V3 training. Compared with accuracy of other traditional methods and deep learning networks, the classification accuracy of the proposed model has been greatly improved and verified. The effectiveness of the model is verified.

Keywords deep learning      transfer learning      convolutional neural network      Inception-V3      remote sensing image classification      scene classification     
:  TP79  
Corresponding Authors: WENG Qian     E-mail:;
Issue Date: 09 October 2020
E-mail this article
E-mail Alert
Articles by authors
Zhiling CAI
Shaozhen YE
Cairen JIAN
Cite this article:   
Zhiling CAI,Qian WENG,Shaozhen YE, et al. Remote sensing image scene classification based on Inception-V3[J]. Remote Sensing for Land & Resources, 2020, 32(3): 80-89.
URL:     OR
Fig.1  Inception-V3 network architecture
Fig.2  Dropout and BN schematics
Fig.3  Freezing and fine-tuning in transfer learning
Fig.4  Sketch of the fine-tuning steps
Fig.5  Flow chart of scene classification for high-resolution remote sensing images
Fig.6  Inception and Inception-L1 training on AID datasets
Fig.7  Inception and Inception-L1 training on NWPU-RESISC45 datasets
Fig.8  Different Dropout rate training on AID datasets
Fig.9  Different Dropout rate training on NWPU-RESISC45 datasets
Inception-L1 test_accuracy
Dropout率0.1 Dropout率0.2 Dropout率0.4
AID 94.30±0.25 94.44±0.23 94.40±0.31
NWPU-RESISC45 93.70±0.28 93.91±0.15 3.95±0.26
Tab.1  Comparison of classification accuracy of different Dropout rates(%)
方法 test_accuracy 方法 test_accuracy
LBP 26.26±0.52 21.74±0.18 SPM+CH 41.27±0.49 41.82±0.21
CH 34.29±0.40 27.52±0.14 VLAD+CH 44.78±0.28 50.57±0.48
SIFT 13.24±0.74 11.48±0.21 AlexNet 86.34±0.43 79.24±0.10
GIST 30.61±0.63 17.88±0.22 VGG-16 86.87±0.41 82.21±0.32
BoVW+CH 47.77±0.52 49.87±0.23 GooLeNet 83.84±0.36 78.47±0.28
IFK+CH 64.83±0.42 66.47±0.27 ResNet50 89.70±1.05 88.35±0.49
LLC+CH 49.36±0.57 46.81±0.30 Inception-V3 94.18±0.40 93.40±0.28
pLSA+CH 42.87±0.54 41.97±0.43 Inception-L1 94.44±0.23 93.95±0.15
Tab.2  Comparison of classification accuracy(%)
类别 准确率 类别 准确率
森林 100 教堂 96
裸地 100 储存罐 96
棒球场 100 飞机场 95
沙滩 100 95
山脉 100 港口 95
稀疏住宅区 100 火车站 95
草地 100 商业区 93
体育场 99 密集住宅区 92
高架桥 99 中型住宅区 92
停车场 99 工业区 91
池塘 98 学校 85
河流 98 景区 82
操场 97 广场 82
沙漠 97 中心区 78
农田 97 公园 76
Tab.3  Classification results of Inception-L1 in AID dataset(%)
Fig.10  Legend of center, church, park and resort
类别 准确率 类别 准确率
灌木丛 100 湖泊 95
圆形农田 99 高速公路 95
小岛 99 95
梯田 99 网球场 95
海冰 99 沙漠 94
棒球内场 98 稀疏住宅区 94
98 火电站 94
高尔夫球场 98 飞机 94
田径场 98 飞机场 94
港口 98 工业区 93
活动房区 98 山脉 93
停车场 98 跑道 92
雪堡 98 矩形农田 91
体育场 97 中型住宅区 91
储存罐 97 铁道 90
环岛 97 河流 90
草地 96 湿地 89
篮球场 96 密集住宅区 88
森林 96 火车站 88
96 商业区 87
立交桥 96 教堂 79
十字路口 95 宫殿 71
沙滩 95
Tab.4  Classification results of Inception-L1 in NWPU-RESISC45 dataset(%)
Fig.11  Legend of church and palace
[1] 崔璐, 张鹏, 车进. 基于深度神经网络的遥感图像分类算法综述[J]. 计算机科学, 2018,45(s1):50-53.
[1] Cui L, Zhang P, Che J. Overview of remote sensing image classification algorithm based on depth neural network[J]. Computer Science, 2018,45(s1):50-53.
[2] 张康, 黑保琴, 李盛阳, 等. 基于CNN模型的遥感图像复杂场景分类[J]. 国土资源遥感, 2018,30(4):49-55.doi: 10.6046/gtzyyg.2018.04.08.
[2] Zhang K, Hei B Q, Li S Y, et al. Classification of complex scene of remote sensing image based on CNN model[J]. Remote Sensing of Land and Resources, 2018,30(4):49-55.doi: 10.6046/gtzyyg.2018.04.08.
[3] Chen C, Zhang B, Su H, et al. Land-use scene classification using multi-scale completed local binary patterns[J]. Signal,Image & Video Processing, 2016,10(4):745-752.
[4] Xia G S, Hu J W, Hu F, et al. AID:A benchmark data set for performance evaluation of aerial scene classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017,55(7):3965-3981.
doi: 10.1109/TGRS.2017.2685945 url:
[5] Swain M J, Ballard D H. Color indexing[J]. International Journal of Computer Vision, 1991,7(1):11-32.
doi: 10.1007/BF00130487 url:
[6] Oliva A, Torralba A. Modeling the shape of the scene:A holistic representation of the spatial envelope[J]. International Journal of Computer Vision, 2001,42(3):145-175.
doi: 10.1023/A:1011139631724 url:
[7] Ojala T, Pietikainen M, Maenpaa T. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns [C]//European Conference on Computer Vision.Springer, 2000: 404-420.
[8] Lowe D G. Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004,60(2):91-110.
doi: 10.1023/B:VISI.0000029664.99615.94 url:
[9] Yang Y, Newsam S. Bag-of-visual-words and spatial extensions for land-use classification [C]//Proceedings of the 18th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems.San Jose:ACM, 2010: 270-279.
[10] Lazebnik S, Schmid C, Ponce J. Beyond bags of features:Spatial pyramid matching for recognizing natural scene categories [C]//2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).New York:IEEE, 2006: 2169-2178.
[11] Wang J, Yang J, Yu K, et al. Locality-constrained linear coding for image classification [C]//2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.San Francisco:IEEE, 2010: 3360-3367.
[12] Bosch A, Zisserman A, Muñoz X. Scene classification via pLSA [C]//European Conference on Computer Vision.Springer, 2006: 517-530.
[13] Perronnin F, Sánchez J, Mensink T. Improving the fisher kernel for large-scale image classification [C]//European Conference on Computer Vision.Springer, 2010: 143-156.
[14] Jegou H, Perronnin F, Douze M, et al. Aggregating local image descriptors into compact codes[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012,34(9):1704-1716.
doi: 10.1109/TPAMI.2011.235 url:
[15] Zou Q, Ni L, Zhang T, et al. Deep learning based feature selection for remote sensing scene classification[J]. IEEE Geoscience and Remote Sensing Letters, 2015,12(11):2321-2325.
doi: 10.1109/LGRS.2015.2475299 url:
[16] 葛芸, 江顺亮, 叶发茂, 等. 聚合CNN特征的遥感图像检索[J]. 国土资源遥感, 2019,31(1):49-57.doi: 10.6046/gtzyyg.2019.01.07.
[16] Ge Y, Jiang S L, Ye F M, et al. Remote sensing image retrieval based on CNN features[J]. Remote Sensing for Land and Resources, 2019,31(1):49-57.doi: 10.6046/gtzyyg.2019.01.07.
[17] Weng Q, Mao Z, Lin J, et al. Land-use classification via extreme learning classifier based on deep convolutional features[J]. IEEE Geoscience and Remote Sensing Letters, 2017,14(5):704-708.
doi: 10.1109/LGRS.2017.2672643 url:
[18] 钱晓亮, 李佳, 程塨, 等. 特征提取策略对高分辨率遥感图像场景分类性能影响的评估[J]. 遥感学报, 2018,22(5):758-776.
[18] Qian X L, Li J, Cheng G, et al. Evaluation of the effect of feature extraction strategy on scene classification performance of high resolution remote sensing image[J]. Journal of Remote Sensing, 2018,22(5):758-776.
[19] 党宇, 张继贤, 邓喀中, 等. 基于深度学习AlexNet的遥感影像地表覆盖分类评价研究[J]. 地球信息科学学报, 2017,19(11):1530-1537.
doi: 10.3724/SP.J.1047.2017.01530 url:
[19] Dang Y, Zhang J X, Deng K Z, et al. Classification and evaluation of surface coverage of remote sensing images based on deep learning AlexNet[J]. Journal of Geoinformatics, 2017,19(11):1530-1537.
[20] Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks [C]//Advances in Neural Information Processing Systems, 2012: 1097-1105.
[21] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[EB/OL].(2014-12-23)[2019-11-06].
[22] Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.IEEE, 2015: 1-9.
[23] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2016: 770-778.
[24] 王鑫, 李可, 徐明君, 等. 改进的基于深度学习的遥感图像分类算法[J]. 计算机应用, 2019,39(2):382-387.
[24] Wang X, Li K, Xu M J, et al. Improved classification algorithm of remote sensing image based on deep learning[J]. Computer Applications, 2019,39(2):382-387.
[25] 刘大伟, 韩玲, 韩晓勇. 基于深度学习的高分辨率遥感影像分类研究[J]. 光学学报, 2016,36(4):306-314.
[25] Liu D W, Han L, Han X Y. Research on high-resolution remote sensing image classification based on deep learning[J]. Acta Optics Sinica, 2016,36(4):306-314.
[26] Szegedy C, Liu W, Jia Y Q, et al. Going deeper with convolutions [C]//Conference on Computer Vision and Pattern Recognition. 2015: 1-9.
[27] Li X, Chen S, Hu X, et al.Understanding the disharmony between dropout and batch normalization by variance shift[EB/OL].(2018 -01-16)[2019-11-06].
[28] 林志玮, 涂伟豪, 黄嘉航, 等. 基于FC-DenseNet的低空航拍光学图像树种识别[J]. 国土资源遥感, 2019,31(3):225-233.doi: 10.6046/gtzyyg.2019.03.28.
[28] Lin Z W, Tu W H, Huang J H, et al. Tree species recognition based on FC-DenseNet in low altitude aerial optical images[J]. Remote Sensing for Land and Resources, 2019,31(3):225-233.doi: 10.6046/gtzyyg.2019.03.28.
[1] XUE Bai, WANG Yizhe, LIU Shuhan, YUE Mingyu, WANG Yiying, ZHAO Shihu. Change detection of high-resolution remote sensing images based on Siamese network[J]. Remote Sensing for Natural Resources, 2022, 34(1): 61-66.
[2] YU Xinli, SONG Yan, YANG Miao, HUANG Lei, ZHANG Yanjie. Multi-model and multi-scale scene recognition of shipbuilding enterprises based on convolutional neural network with spatial constraints[J]. Remote Sensing for Natural Resources, 2021, 33(4): 72-81.
[3] GUO Xiaozheng, YAO Yunjun, JIA Kun, ZHANG Xiaotong, ZHAO Xiang. Information extraction of Mars dunes based on U-Net[J]. Remote Sensing for Natural Resources, 2021, 33(4): 130-135.
[4] FENG Dongdong, ZHANG Zhihua, SHI Haoyue. Fine extraction of urban villages in provincial capitals based on multivariate data[J]. Remote Sensing for Natural Resources, 2021, 33(3): 272-278.
[5] WU Yu, ZHANG Jun, LI Yixu, HUANG Kangyu. Research on building cluster identification based on improved U-Net[J]. Remote Sensing for Land & Resources, 2021, 33(2): 48-54.
[6] LU Qi, QIN Jun, YAO Xuedong, WU Yanlan, ZHU Haochen. Buildings extraction of GF-2 remote sensing image based on multi-layer perception network[J]. Remote Sensing for Land & Resources, 2021, 33(2): 75-84.
[7] AN Jianjian, MENG Qingyan, HU Die, HU Xinli, YANG Jian, YANG Tianliang. The detection and determination of the working state of cooling tower in the thermal power plant based on Faster R-CNN[J]. Remote Sensing for Land & Resources, 2021, 33(2): 93-99.
[8] LIU Zhao, ZHAO Tong, LIAO Feifan, LI Shuai, LI Haiyang. Research and comparative analysis on urban built-up area extraction methods from high-resolution remote sensing image based on semantic segmentation network[J]. Remote Sensing for Land & Resources, 2021, 33(1): 45-53.
[9] CAI Xiang, LI Qi, LUO Yan, QI Jiandong. Surface features extraction of mining area image based on object-oriented and deep-learning method[J]. Remote Sensing for Land & Resources, 2021, 33(1): 63-71.
[10] QIU Yifan, CHAI Dengfeng. A deep learning method for Landsat image cloud detection without manually labeled data[J]. Remote Sensing for Land & Resources, 2021, 33(1): 102-107.
[11] ZHENG Zhiteng, FAN Haisheng, WANG Jie, WU Yanlan, WANG Biao, HUANG Tengjie. An improved double-branch network method for intelligently extracting marine cage culture area[J]. Remote Sensing for Land & Resources, 2020, 32(4): 120-129.
[12] DU Fangzhou, SHI Yuli, SHENG Xia. Research on downscaling of TRMM precipitation products based on deep learning: Exemplified by northeast China[J]. Remote Sensing for Land & Resources, 2020, 32(4): 145-153.
[13] PEI Chan, LIAO Tiejun. Multi-scale architecture search method for remote sensing object detection[J]. Remote Sensing for Land & Resources, 2020, 32(4): 53-60.
[14] WEI Hongyu, ZHAO Yindi, DONG Jihong. Cooling tower detection based on the improved RetinaNet[J]. Remote Sensing for Land & Resources, 2020, 32(4): 68-73.
[15] LIU Zhao, LIAO Feifan, ZHAO Tong. Remote sensing image urban built-up area extraction and optimization method based on PSPNet[J]. Remote Sensing for Land & Resources, 2020, 32(4): 84-89.
Full text



Copyright © 2017 Remote Sensing for Natural Resources
Support by Beijing Magtech