Corpus ID: 54101493

ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness

@article{Geirhos2019ImageNettrainedCA,
  title={ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness},
  author={Robert Geirhos and Patricia Rubisch and Claudio Michaelis and M. Bethge and Felix Wichmann and W. Brendel},
  journal={ArXiv},
  year={2019},
  volume={abs/1811.12231}
}
Convolutional Neural Networks (CNNs) are commonly thought to recognise objects by learning increasingly complex representations of object shapes. [...] Key Method We then demonstrate that the same standard architecture (ResNet-50) that learns a texture-based representation on ImageNet is able to learn a shape-based representation instead when trained on "Stylized-ImageNet", a stylized version of ImageNet. This provides a much better fit for human behavioural performance in our well-controlled psychophysical lab…Expand
Shape or Texture: Understanding Discriminative Features in CNNs
Exploring the Origins and Prevalence of Texture Bias in Convolutional Neural Networks
Curriculum By Texture
Hiding a plane with a pixel: examining shape-bias in CNNs and the benefit of building in biological constraints
Assessing The Importance Of Colours For CNNs In Object Recognition
CognitiveCNN: Mimicking Human Cognitive Models to resolve Texture-Shape Bias
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 38 REFERENCES
Assessing Shape Bias Property of Convolutional Neural Networks
Correspondence of Deep Neural Networks and the Brain for Visual Textures
Deep Neural Networks as a Computational Model for Human Shape Sensitivity
The role of textural statistics vs. outer contours in deep CNN and neural responses to objects
Generalisation in humans and deep neural networks
Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet
Cognitive Psychology for Deep Neural Networks: A Shape Bias Case Study
Deep Neural Networks Rival the Representation of Primate IT Cortex for Core Visual Object Recognition
...
1
2
3
4
...