Recognizing Instagram Filtered Images with Feature De-stylization

by   Zhe Wu, et al.

Deep neural networks have been shown to suffer from poor generalization when small perturbations are added (like Gaussian noise), yet little work has been done to evaluate their robustness to more natural image transformations like photo filters. This paper presents a study on how popular pretrained models are affected by commonly used Instagram filters. To this end, we introduce ImageNet-Instagram, a filtered version of ImageNet, where 20 popular Instagram filters are applied to each image in ImageNet. Our analysis suggests that simple structure preserving filters which only alter the global appearance of an image can lead to large differences in the convolutional feature space. To improve generalization, we introduce a lightweight de-stylization module that predicts parameters used for scaling and shifting feature maps to "undo" the changes incurred by filters, inverting the process of style transfer tasks. We further demonstrate the module can be readily plugged into modern CNN architectures together with skip connections. We conduct extensive studies on ImageNet-Instagram, and show quantitatively and qualitatively, that the proposed module, among other things, can effectively improve generalization by simply learning normalization parameters without retraining the entire network, thus recovering the alterations in the feature space caused by the filters.


page 1

page 4

page 7


Deep Feature Space Trojan Attack of Neural Networks by Controlled Detoxification

Trojan (backdoor) attack is a form of adversarial attack on deep neural ...

Inability of spatial transformations of CNN feature maps to support invariant recognition

A large number of deep learning architectures use spatial transformation...

Rethinking and Improving the Robustness of Image Style Transfer

Extensive research in neural style transfer methods has shown that the c...

Performance degradation of ImageNet trained models by simple image transformations

ImageNet trained PyTorch models are generally preferred as the off-the-s...

Standardised convolutional filtering for radiomics

The Image Biomarker Standardisation Initiative (IBSI) aims to improve re...

Gabor is Enough: Interpretable Deep Denoising with a Gabor Synthesis Dictionary Prior

Image processing neural networks, natural and artificial, have a long hi...

Please sign up or login with your details

Forgot password? Click here to reset