Skip to main content
SpringerLink
Log in

Visualizing Deep Convolutional Neural Networks Using Natural Pre-images

  • Published:
International Journal of Computer Vision Aims and scope Submit manuscript

Abstract

Image representations, from SIFT and bag of visual words to convolutional neural networks (CNNs) are a crucial component of almost all computer vision systems. However, our understanding of them remains limited. In this paper we study several landmark representations, both shallow and deep, by a number of complementary visualization techniques. These visualizations are based on the concept of “natural pre-image”, namely a natural-looking image whose representation has some notable property. We study in particular three such visualizations: inversion, in which the aim is to reconstruct an image from its representation, activation maximization, in which we search for patterns that maximally stimulate a representation component, and caricaturization, in which the visual patterns that a representation detects in an image are exaggerated. We pose these as a regularized energy-minimization framework and demonstrate its generality and effectiveness. In particular, we show that this method can invert representations such as HOG more accurately than recent alternatives while being applicable to CNNs too. Among our findings, we show that several layers in CNNs retain photographically accurate information about the image, with different degrees of geometric and photometric invariance.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16
Fig. 17
Fig. 18
Fig. 19
Fig. 20
Fig. 21

Similar content being viewed by others

Notes

  1. It is referred to as TV norm in Mahendran and Vedaldi (2015) but for \(\beta =2\) this is actually the quadratic norm.

  2. In the following, the image \(\mathbf {x}\) is assumed to have null mean, as required by most CNN implementations.

  3. This requires addressing a few more subtleties. Please see files dsift_net.m and hog_net.m for details.

References

  • Bishop, C. M. (1995). Neural networks for pattern recognition. Oxford: Clarendon Press.

    MATH  Google Scholar 

  • Chatfield, K., Simonyan, K., Vedaldi, A., & Zisserman, A. (2014). Return of the devil in the details: Delving deep into convolutional nets. In Proceedings of BMVC.

  • Chen, Y., Ranftl, R., & Pock, T. (2014). A bi-level view of inpainting-based image compression. In Proceedings of computer vision winter workshop.

  • Csurka, G., Dance, C. R., Dan, L., Willamowski, J., & Bray, C. (2004). Visual categorization with bags of keypoints. In Proceedings of ECCV workshop on statistical learning in computer vision.

  • Dalal, N., & Triggs, B. (2005). Histograms of oriented gradients for human detection. In CVPR.

  • d’Angelo, E., Alahi, A., & Vandergheynst, P. (2012). Beyond bits: Reconstructing images from local binary descriptors. In ICPR (pp. 935–938).

  • Dosovitskiy, A., & Brox, T. (2015). Inverting convolutional networks with convolutional networks. CoRR. arXiv:1506.02753.

  • Duchi, J., Hazan, E., & Singer, Y. (2011). Adaptive subgradient methods for online learning and stochastic optimization. Journal of Machine Learninig Research, 12, 2121–2159.

    MathSciNet  MATH  Google Scholar 

  • Erhan, D., Bengio, Y., Courville, A., & Vincent, P. (2009). Visualizing higher-layer features of a deep network. Technical report (Vol. 1341). Montreal: University of Montreal.

  • Everingham, M., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2010). The PASCAL visual object classes challenge 2010 (VOC2010) results. http://host.robots.ox.ac.uk/pascal/VOC/voc2010/results/index.html. Accessed 10 Apr 2016.

  • Felzenszwalb, P. F., Girshick, R. B., McAllester, D., & Ramanan, D. (2010). Object detection with discriminatively trained part based models. IEEE Transactions on Pattern Analysis and Machine Intelligence, 32(9), 1627–1645.

    Article  Google Scholar 

  • Gatys, L. A., Ecker, A. S., & Bethge, M. (2015). A neural algorithm of artistic style. CoRR. arXiv:1508.06576.

  • Gatys, L. A., Ecker, A. S., & Bethge, M. (2015). Texture synthesis and the controlled generation of natural stimuli using convolutional neural networks. In NIPS proceedings.

  • Girshick, R. B., Felzenszwalb, P. F., & McAllester, D. (2010). Discriminatively trained deformable part models, release 5. http://people.cs.uchicago.edu/~rbg/latent-release5/. Accessed 10 April 2016.

  • Hinton, G. E., & Salakhutdinov, R. R. (2006). Reducing the dimensionality of data with neural networks. Science, 313, 5786.

    Article  MathSciNet  MATH  Google Scholar 

  • Huang, W. J., & Mumford, D. (1999). Statistics of natural images and models. In Proceedings of CVPR.

  • Jégou, H., Douze, M., Schmid, C., & Pérez, P. (2010). Aggregating local descriptors into a compact image representation. In CVPR.

  • Jensen, C. A., Reed, R. D., Marks, R. J., El-Sharkawi, M., Jung, J. B., Miyamoto, R., et al. (1999). Inversion of feedforward neural networks: Algorithms and applications. Proceedings of the IEEE, 87, 9.

    Article  Google Scholar 

  • Jia, Y. (2013). Caffe: An open source convolutional architecture for fast feature embedding. http://caffe.berkeleyvision.org/. Accessed 14 Nov 2014.

  • Julesz, B. (1981). Textons, the elements of texture perception, and their interactions. Nature, 290(5802), 91–97.

    Article  Google Scholar 

  • Kato, H., & Harada, T. (2014). Image reconstruction from bag-of-visual-words. In CVPR.

  • Krishnan, D., & Fergus, R. (2009). Fast image deconvolution using hyper-laplacian priors. In NIPS.

  • Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In NIPS.

  • Lee, S., & Kil, R. M. (1994). Inverse mapping of continuous functions using local and global information. IEEE Transactions on Neural Networks, 5, 3.

    Article  Google Scholar 

  • Leung, T., & Malik, J. (2001). Representing and recognizing the visual appearance of materials using three-dimensional textons. IJCV, 43, 1.

    Article  MATH  Google Scholar 

  • Linden, A., & Kindermann, J. (1989). Inversion of multilayer nets. In Proceedings of international conference on neural networks.

  • Lowe, D. G. (1999). Object recognition from local scale-invariant features. In ICCV.

  • Lowe, D. G. (2004). Distinctive image features from scale-invariant keypoints. IJCV, 2(60), 91–110.

    Article  Google Scholar 

  • Lu, B. L., Kita, H., & Nishikawa, Y. (1999). Inverting feedforward neural networks using linear and nonlinear programming. IEEE Transactions on Neural Networks, 10, 6.

    Article  Google Scholar 

  • Mahendran, A., & Vedaldi, A. (2015). Understanding deep image representations by inverting them. In Proceedings of CVPR.

  • Mordvintsev, A., Olah, C., & Tyka, M. (2015). Inceptionism: Going deeper into neural networks. http://googleresearch.blogspot.co.uk/2015/06/inceptionism-going-deeper-into-neural.html. Accessed 17 Oct 2015.

  • Nguyen, A., Yosinski, J., & Clune, J. (2015). Deep neural networks are easily fooled: High confidence predictions for unrecognizable images. In Proceedings of CVPR.

  • Nowak, E., Jurie, F., & Triggs, B. (2006). Sampling strategies for bag-of-features image classification. In ECCV.

  • Perronnin, F., & Dance, C. (2006). Fisher kernels on visual vocabularies for image categorizaton. In CVPR.

  • Portilla, J., & Simoncelli, E. P. (2000). A parametric texture model based on joint statistics of complex wavelet coefficients. IJCV, 40, 49–70.

    Article  MATH  Google Scholar 

  • Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., et al. (2015). ImageNet large scale visual recognition challenge. IJCV, 115(3), 211–252. doi:10.1007/s11263-015-0816-y.

    Article  MathSciNet  Google Scholar 

  • Sermanet, P., Eigen, D., Zhang, X., Mathieu, M., Fergus, R., & LeCun, Y. (2014). Overfeat: Integrated recognition, localization and detection using convolutional networks. In CoRR. arXiv:1312.6229.

  • Simonyan, K., Vedaldi, A., & Zisserman, A. (2014). Deep inside convolutional networks: Visualising image classification models and saliency maps. In Proceedings of ICLR.

  • Sivic, J., & Zisserman, A. (2003). Video Google: A text retrieval approach to object matching in videos. In ICCV

  • Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I. J., et al. (2014). Intriguing properties of neural networks. In Proceedings of ICLR.

  • Tatu, A., Lauze, F., Nielsen, M., & Kimia, B. (2011). Exploring the representation capabilities of the HOG descriptor. In ICCV workshop.

  • Várkonyi-Kóczy, A. R., & Rövid, A. (2005). Observer based iterative neural network model inversion. In IEEE international conference on fuzzy systems.

  • Vedaldi, A. (2007). An open implementation of the SIFT detector and descriptor. Technical report 070012. Los Angeles: UCLA CSD.

  • Vedaldi, A., & Lenc, K. (2014). MatConvNet: CNNs for MATLAB. http://www.vlfeat.org/matconvnet/. Accessed 17 Oct 2015.

  • Vondrick, C., Khosla, A., Malisiewicz, T., & Torralba, A. (2013). HOGgles: Visualizing object detection features. In ICCV.

  • Wang, J., Yang, J., Yu, K., Lv, F., Huang, T., & Gong, Y. (2010). Locality-constrained linear coding for image classification. In CVPR.

  • Weinzaepfel, P., Jégou, H., & Pérez, P. (2011). Reconstructing an image from its local descriptors. In CVPR.

  • Williams, R. J. (1986). Inverting a connectionist network mapping by back-propagation of error. In Proceedings of CogSci.

  • Yang, J., Yu, K., & Huang, T. (2010). Supervised translation-invariant sparse coding. In CVPR.

  • Yosinksi, J., Clune, J., Nguyen, A., Fuchs, T., & Lipson, H. (2015). Understanding neural networks through deep visualization. In Proceedings of ICML workshop.

  • Zeiler, M. D. (2012). ADADELTA: An adaptive learning rate method. CoRR. arXiv:1212.5701.

  • Zeiler, M. D., & Fergus, R. (2014). Visualizing and understanding convolutional networks. In ECCV.

  • Zhou, X., Yu, K., Zhang, T., & Huang, T. S. (2010). Image classification using super-vector coding of local image descriptors. In ECCV.

  • Zhu, C., Byrd, R. H., Lu, P., & Nocedal, J. (1997). Algorithm 778: L-bfgs-b: Fortran subroutines for large-scale bound-constrained optimization. ACM Transactions on Mathematical Software (TOMS), 23(4), 550–560.

    Article  MathSciNet  MATH  Google Scholar 

  • Zhu, S. C., Wu, Y., & Mumford, D. (1998). Filters, random fields and maximum entropy (FRAME): Towards a unified theory for texture modeling. IJCV, 27(2), 4.

    Article  Google Scholar 

Download references

Acknowledgments

We gratefully acknowledge the support of the ERC StG IDIU for Andrea Vedaldi and of BP for Aravindh Mahendran.

Author information

Authors and Affiliations

  1. University of Oxford, Oxford, UK

    Aravindh Mahendran & Andrea Vedaldi

Authors
  1. Aravindh Mahendran
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Andrea Vedaldi
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Aravindh Mahendran.

Additional information

Communicated by Cordelia Schmid.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Mahendran, A., Vedaldi, A. Visualizing Deep Convolutional Neural Networks Using Natural Pre-images. Int J Comput Vis 120, 233–255 (2016). https://doi.org/10.1007/s11263-016-0911-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11263-016-0911-8

Keywords

Search

Navigation