A study on the invariance in security whatever the dimension of images for the steganalysis by deep-learning

Kévin Planolles, Marc Chaumont, Frédéric Comby
2/22/2023
cs.CVcs.CR

No official code found. Best implementation: Kevin-Planolles/steganalysis_with_CNN_dilated-Yedroudj-Net (3 stars, last commit 2023-02-22).

Abstract

In this paper, we study the performance invariance of convolutional neural networks when confronted with variable image sizes in the context of a more "wild steganalysis". First, we propose two algorithms and definitions for a fine experimental protocol with datasets owning "similar difficulty" and "similar security". The "smart crop 2" algorithm allows the introduction of the Nearly Nested Image Datasets (NNID) that ensure "a similar difficulty" between various datasets, and a dichotomous research algorithm allows a "similar security". Second, we show that invariance does not exist in state-of-the-art architectures. We also exhibit a difference in behavior depending on whether we test on images larger or smaller than the training images. Finally, based on the experiments, we propose to use the dilated convolution which leads to an improvement of a state-of-the-art architecture.

View on arXivView PDF

Code Implementations(1)

Kevin-Planolles/steganalysis_with_CNN_dilated-Yedroudj-NetCommunity100%

A variation on the Yedroudj-Net network for deep learning based steganalysis.

30PythonOct 13, 20223 years ago

Similar Papers

Why do deep convolutional networks generalize so poorly to small image transformations?

83% match1 repo112018

On the Robustness of Kolmogorov-Arnold Networks: An Adversarial Perspective

82% match2024

Exploring the Space of Adversarial Images

82% match1 repo702015

Deep Convolutional Neural Network to Detect J-UNIWARD

81% match1 repo302017

Hierarchical Invariance for Robust and Interpretable Vision Tasks at Larger Scales

81% match2024

Not So Robust After All: Evaluating the Robustness of Deep Neural Networks to Unseen Adversarial Attacks

81% match2023

Cite this paper

@article{planolles2023study,
  title  = {A study on the invariance in security whatever the dimension of images for the steganalysis by deep-learning},
  author = {Kévin Planolles and Marc Chaumont and Frédéric Comby},
  year   = {2023},
  eprint = {2302.11527},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2302.11527v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2302.11527.svg)](https://codeofpaper.com/paper/2302.11527)

Discussion