Real-world actor-based image steganalysis via classifier inconsistency detection
Daniel Lerch-Hostalot, David Megías
TL;DR
This work tackles real-world actor-based image steganalysis under Cover Source Mismatch (CSM) by integrating Detection of Classifier Inconsistencies (DCI) with EfficientNet-based feature extraction and a Gradient Boosting Machine. Two EfficientNet models are trained per stegosystem to enable primary and secondary DCI predictions, and a GBM uses these predictions to infer actor innocence, guilt, or discard actors due to CSM, with a 0.7 threshold guiding discards. The approach demonstrates robust accuracy above 80% even when CSM is as high as 100% in both spatial and JPEG domains, outperforming a baseline that relies only on stego-ratios. This yields a practical, scalable solution for real-world steganalysis where source diversity and unknown payloads challenge traditional classifiers, and it provides clear mechanisms to identify and discard high-CSM cases.
Abstract
In this paper, we propose a robust method for detecting guilty actors in image steganography while effectively addressing the Cover Source Mismatch (CSM) problem, which arises when classifying images from one source using a classifier trained on images from another source. Designed for an actor-based scenario, our method combines the use of Detection of Classifier Inconsistencies (DCI) prediction with EfficientNet neural networks for feature extraction, and a Gradient Boosting Machine for the final classification. The proposed approach successfully determines whether an actor is innocent or guilty, or if they should be discarded due to excessive CSM. We show that the method remains reliable even in scenarios with high CSM, consistently achieving accuracy above 80% and outperforming the baseline method. This novel approach contributes to the field of steganalysis by offering a practical and efficient solution for handling CSM and detecting guilty actors in real-world applications.