Advertisement

Hardware Implementation of Convolutional Neural Network-Based Remote Sensing Image Classification Method

  • Lei Chen
  • Xin Wei
  • Wenchao Liu
  • He Chen
  • Liang ChenEmail author
Conference paper
Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 516)

Abstract

The convolutional neural networks have achieved very good results in the field of remote sensing image classification and recognition. However, the cost of huge computational complexity with the significant accuracy improvement of CNNs makes a huge challenge to hardware implementation. A promising solution is FPGA due to it supports parallel computing with low power consumption. In this paper, LeNet-5-based remote sensing image classification method is implemented on FPGA. The test images with a size of 126 × 126 are transformed to the system from PC by serial port. The classification accuracy is 98.18% tested on the designed system, which is the same as that on PC. In the term of efficiency, the designed system runs 2.29 ms per image, which satisfies the real-time requirements.

Keywords

CNN Remote sensing image FPGA Classification 

Notes

Acknowledgements

This work was supported by the Chang Jiang Scholars Programmed under Grant T2012122 and the Youth Science and Technology Innovation Leader of National Innovation Talent Promotion Program under Grant No. 2013RA2034.

References

  1. 1.
    Garea AS, Heras DB, Argüello F. Caffe CNN-based classification of hyperspectral images on GPU. J Supercomputing. 2018;3:1–13.Google Scholar
  2. 2.
    Hu F, Xia GS, Hu J, et al. Transferring deep convolutional neural networks for the scene classification of high-resolution remote sensing imagery. Remote Sens. 2015;7(11):14680–707.CrossRefGoogle Scholar
  3. 3.
    Liu W, Ma L, Chen H. Arbitrary-oriented ship detection framework in optical remote-sensing images. IEEE Geosci Remote Sens Lett. 2018;99:1–5.Google Scholar
  4. 4.
    Lecun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition. Proc IEEE. 1998;86(11):2278–324.CrossRefGoogle Scholar
  5. 5.
    Qiu J, Wang J, Yao S, et al. Going deeper with embedded FPGA platform for convolutional neural network. In: Acm/sigda international symposium on field-programmable gate arrays. ACM; 2016. p. 26–35.Google Scholar
  6. 6.
    Ren S, He K, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks. In: International conference on neural information processing systems. MIT Press; 2015. p. 91–9.Google Scholar
  7. 7.
    Vedaldi A, Lenc K. MatConvNet: convolutional neural networks for MATLAB; 2015:689–92.Google Scholar
  8. 8.
    Yang Y, Zhuang Y, Bi F, et al. M-FCN: effective fully convolutional network-based airplane detection framework. IEEE Geosci Remote Sens Lett. 2017;14(8):1293–7.CrossRefGoogle Scholar
  9. 9.
    Zhang C, Li P, Sun G, et al. Optimizing FPGA-based accelerator design for deep convolutional neural networks. In: Acm/sigda international symposium on field-programmable gate arrays. ACM; 2015. p. 161–70.Google Scholar
  10. 10.
    Zhang C, Wu D, Sun J, et al. Energy-efficient CNN implementation on a deeply pipelined FPGA cluster. In: International symposium on low power electronics and design. ACM; 2016. p. 326–31.Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2020

Authors and Affiliations

  • Lei Chen
    • 1
    • 2
  • Xin Wei
    • 1
    • 2
  • Wenchao Liu
    • 1
    • 2
  • He Chen
    • 1
    • 2
  • Liang Chen
    • 1
    • 2
    Email author
  1. 1.Radar Research Lab, School of Information and ElectronicsBeijing Institute of TechnologyBeijingChina
  2. 2.Beijing Key Laboratory of Embedded Real-Time Information Processing TechnologyBeijing Institute of TechnologyBeijingChina

Personalised recommendations