Laijin Meng

The rapid advancement of facial manipulation technologies demands detection systems that can generalize to novel forgery techniques. We identify that Standard Learning, reliant on static datasets and uniform sampling, detrimentally biases models towards specific patterns tied to individual generation techniques, hindering their ability to learn general features. To overcome this, we introduce Adaptive Learning (AL) for face forgery detection, a cyclical framework that simultaneously refines both the detector model and the training data through dynamic sample selection and model optimization. AL’s efficacy hinges on identifying samples rich in generalizable forgery clues. Thus, we propose Augmentation Robustness Validation (ARV) as AL’s core purification engine. ARV exploits the stability of predictions across diverse semantic-preserving augmentations as a reliable proxy for general feature presence: samples that exhibit invariant predictions inherently contain robust manipulation traces. Integrating ARV with AL yields Adaptive Learning with Augmentation Robustness Validation (ALarv). ALarv strategically prioritizes stability-verified samples during iterative training cycles, progressively enhancing the model’s focus on transferable forensic features. Inspired by the architectural advantages of ConvNeXt, we incorporate it into ALarv, forming an effective method, ALarv-ConvNeXt. Extensive experiments demonstrate ALarv-ConvNeXt’s superior generalization performance, including emerging diffusion-based synthetic faces.

Coverless image steganography (CIS) can completely resist steganalysis algorithms as it does not modify the carriers. However, most existing CIS methods overlook the diversity of generated hash sequences, which directly impacts the practical maximum capacity and difficulty of constructing the index structure. Moreover, CIS methods typically consider robustness only under known parameters. In security, some methods require the transmission of auxiliary information along with stego images, while others reuse fixed images in the dataset, which raises potential suspicion from attackers. To address the above issues, a robust CIS method towards resisting black-box attacks is proposed. The normalized average of keypoint descriptor features is calculated by Scale-Invariant Feature Transform (SIFT) to construct the hierarchical coverless image dataset (CID). Subsequently, the mapping table between hash sequences and images in CID is established to accomplish secret information hiding and extraction. Experimental results demonstrate that the proposed CIS method achieves a higher capacity and stronger robustness than the state-of-the-art CIS methods. Notably, it is the first time to consider robustness against black-box and mixed attacks in CIS methods, making it more applicable to real-world scenarios.

Compared with traditional video steganography, coverless video steganography (CVS) can completely avoid being detected by steganalysis algorithms. Recently, the study of CVS has developed rapidly. However, it is still far from the theoretical maximum values in capacity, i.e., the theoretical limit is $2^\ell$ for a hash sequence length of $\ell$. Besides, most existing CVS methods have only considered limited types of video attacks in robustness. In this paper, a novel coverless video steganography based on two-level discrete cosine transform (DCT) features is proposed. First, pre-processing is accomplished on the public video datasets. Then, two-level DCT features are calculated and the Coverless Video Database (CVD) is constructed by the K-means++ clustering algorithm. After that, the mapping table is established to map the secret segments to the CVD. Finally, each secret segment corresponds to a video sequence in the CVD by the mapping table to complete the process of information embedding and extraction. The proposed method first evaluates the robustness against the frame swapping attack, which is a common video attack. Experimental results show that the proposed method can achieve the theoretical maximum value in effective capacity and better robustness compared to the state-of-the-art works.

Compared with traditional modification image steganography, coverless image steganography can resist the detection of steganalysis algorithms relying on no modification to the carriers. Previous works have made great efforts to improve the robustness against image attacks. However, the robustness of resisting geometric attacks performs not that well. After studying the general flow of the coverless image steganography, we find out that the receiver always needs to generate or map the hash sequences directly from the received images, which causes a significantly negative impact for extracting correct secret information because these received images might be attacked. Inspired by this finding, we surprisingly explore a common way to solve the problem by proposing a universal restoration framework for the attacked images. The most important module of the framework, the restoration module, contains two main parts, i.e., the classification sub-module and the attack restoration sub-module. The attacked images at the receiving end are first sent to a classification sub-module to estimate the type of the attack. Then, the corresponding attack restoration sub-module is utilized to repair the attacked images to improve the robustness. Experimental results show that the robustness of the existing coverless image steganography methods have been greatly improved after using the proposed framework without introducing extra security issues.