Tackle CSM in JPEG Steganalysis with Data Adaptation

TL;DR

This paper introduces TADA, a data adaptation framework that improves JPEG steganalysis robustness against cover source mismatch by emulating unknown image processing pipelines from small unlabeled datasets.

Contribution

TADA is a novel approach that learns to emulate unknown image processing pipelines, enhancing steganalysis performance in realistic, unseen scenarios.

Findings

TADA significantly improves robustness to cover source mismatch.

TADA outperforms baseline methods in operational generalization.

The framework effectively emulates unknown processing pipelines.

Abstract

Steganalysis models excel on benchmark datasets but struggle in the wild when analyzed images are produced by a processing pipeline unseen during training. This problem known as Cover Source Mismatch (CSM) is particularly hard in realistic settings where practitioners (1) have access to only a small, unlabeled dataset, (2) are unsure of the processing techniques applied to these images, and (3) lack information on the proportion of covers and stegos in that set. To answer this challenge, we introduce TADA (Target Alignment through Data Adaptation), a framework learning to emulate the unknown processing pipeline from a small unlabeled target set. This architecture is trained with a loss combining residual covariance alignment, residual distribution matching, and a $\ell^2$ loss constraining the emulator to produce realistic images. Across toy and operational targets, TADA yields substantial gains in robustness to CSM and improves operational generalization compared to strong holistic and atomistic baselines. Additional resources are available at this link: https://github.com/RonyAbecidan/TADA