Fast FFT-Based Inference in 3D Convolutional Neural Networks

  • Bo XieEmail author
  • Guidong Zhang
  • Yongjun Shen
  • Shun Liu
  • Yabin Ge
Conference paper
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 773)


Recognizing real world objects based on their 3D shapes is an important problem in robotics, computational medicine, and the internet of things (IoT) applications. In the recent years, deep learning has emerged as the foremost tool for a wide range of recognition and classification problems. However, the main problem of convolutional neural networks, which are the primary deep learning systems for such tasks, lies in the high computational cost required to train and use them, even for 2D problems. For 3D problems, the problem becomes even more pressing, and requires new methods to keep up with the further increase in computational cost. One such method is the use of Fast Fourier Transforms to reduce the computational cost by performing convolution operations in the Fourier domain. Recently, this method has seen widespread use for 2D problems. In this paper, we implement and test the method for 2D and 3D object recognition problems and compare it to the traditional convolution methods. We test our network on the ShapeNet 3D object library, achieving superior performance without any loss in accuracy compared to conventional methods.


  1. 1.
    Krizhevsky, A., Sutskever, I., Hinton, G.E.: ImageNet classification with deep convolutional neural networks. In: Advances in Neural Information Processing Systems, pp. 1097–1105 (2012)Google Scholar
  2. 2.
    Fan, J., Xu, W., Wu, Y., Gong, Y.: Human tracking using convolutional neural networks. IEEE Trans. Neural Netw. 21(10), 1610–1623 (2010)CrossRefGoogle Scholar
  3. 3.
    Redmon, J., Divvala, S., Girshick, R., Farhadi, A.: You only look once: unified, real-time object detection, arXiv preprint arXiv:1506.02640
  4. 4.
    Xiang, Y., Kim, W., Chen, W., Ji, J., Choy, C., Su, H., Mottaghi, R., Guibas, L., Savarese, S.: ObjectNet3d: a large scale database for 3D object recognition. In: European Conference on Computer Vision, pp. 160–176. Springer, (2016)CrossRefGoogle Scholar
  5. 5.
    Chang, A.X., Funkhouser, T., Guibas, L., Hanrahan, P., Huang, Q., Li, Z., Savarese, S., Savva, M., Song, S., Su, H., et al.: ShapeNet: an information-rich 3D model repository, arXiv preprint arXiv:1512.03012
  6. 6.
    Maturana, D., Scherer, S.: VoxNet: a 3D convolutional neural network for real-time object recognition. In: 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 922–928. IEEE (2015)Google Scholar
  7. 7.
    Mathieu, M., Henaff, M., LeCun, Y.: Fast training of convolutional networks through FFTs, arXiv preprint arXiv:1312.5851
  8. 8.
    Han, S., Mao, H., Dally, W.J.: A deep neural network compression pipeline: pruning, quantization, huffman encoding, arXiv preprint arXiv:1510.00149
  9. 9.
    Chen, W., Wilson, J.T., Tyree, S., Weinberger, K.Q., Chen, Y.: Compressing neural networks with the hashing trick, CoRR, abs/1504.04788Google Scholar
  10. 10.
    Wu, J., Leng, C., Wang, Y., Hu, Q., Cheng, J.: Quantized convolutional neural networks for mobile devices, arXiv preprint arXiv:1512.06473
  11. 11.
    Lin, D.D., Talathi, S.S., Annapureddy, V.S.: Fixed point quantization of deep convolutional networks, arXiv preprint arXiv:1511.06393
  12. 12.
    Gupta, S., Agrawal, A., Gopalakrishnan, K., Narayanan, P.: Deep learning with limited numerical precision, CoRR, abs/1502.02551 392Google Scholar
  13. 13.
    Rastegari, M., Ordonez, V., Redmon, J., Farhadi, A.: XNOR-Net: ImageNet classification using binary convolutional neural networks, arXiv preprint arXiv:1603.05279
  14. 14.
    Cong, J., Xiao, B.: Minimizing computation in convolutional neural networks. In: International Conference on Artificial Neural Networks, pp. 281–290. Springer (2014)Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2019

Authors and Affiliations

  • Bo Xie
    • 1
    Email author
  • Guidong Zhang
    • 1
  • Yongjun Shen
    • 1
  • Shun Liu
    • 1
  • Yabin Ge
    • 1
  1. 1.Lanzhou UniversityLanzhouChina

Personalised recommendations