DINO-MAC is a simple yet effective model for robust deepfake detection under degraded conditions and is the first-place winner of the “NTIRE 2026 - Robust Deepfake Detection Challenge @ CVPR 2026”, significantly outperforms all competitors.
An ensemble-based approach, namely the X-Iv2 Ensemble approach, merging Inception ResNet v2 and Xception Net based on their complementary architectures to enhance feature extraction and classification is introduced, integrating Explainable AI (XAI) using Integrated Gradients to interpret decision-making processes.
P. P. Sudharsana, R. Rajalaxmi· Intelligent Data Analysis· 0 citations
Deepfakes pose growing risks to information integrity, yet many detectors perform well only on the datasets they were trained on and remain opaque to human analysts. A robust, explainable detection framework is presented that combines a CNN backbone for extracting spatial artifacts with an LSTM module for modeling temporal inconsistencies across frames. To make decisions auditable, the architecture incorporates Grad-CAM for spatial heatmaps, SHAP for quantitative feature attribution, and LIME for local surrogate explanations. The system was trained primarily on FaceForensics++ with stratified sampling and augmentation to reduce dataset bias and evaluated on multiple external benchmarks to assess cross-domain generalization. Experimental results show strong detection metrics, such as accuracy of 96.3%, precision of 95.8%, recall of 96.7%, and an F1-score of 96.2%, along with robust performance under JPEG compression, Gaussian noise, and FGSM adversarial attacks. By coupling high detection accuracy with transparent explanations, the proposed approach enhances forensic decision support and increases practical readiness for content verification systems.
Lastone Banda, Esther J.· International Journal of Dat...· 0 citations
The rapid advancement of generative AI has made high-quality deepfakes increasingly accessible, raising serious concerns regarding misinformation and digital trust. Consequently, developing reliable deepfake detection systems has become a critical task in multimedia forensics. However, existing methods often face a severe trade-off between cross-dataset generalization and computational efficiency. They heavily rely on large-scale backbones, which limits their deployment on resource-constrained edge devices. To bridge this gap, we propose HiDD-Net, a hierarchical dual-domain distillation framework. Specifically, the teacher network integrates multi-granularity spatial views (global, local, and facial) via a cross-attention mechanism to explicitly align fine-grained anomalies with global context. Simultaneously, it captures subtle upsampling spectral artifacts using a three-level discrete wavelet transform (DWT) and Spatial Rich Model (SRM) features. Subsequently, we distill these fused spatial-frequency representations into a compact student model (MobileNetV3) by enforcing feature alignment in a shared semantic space. Extensive evaluations on DeepfakeBench demonstrate that the student achieves a competitive cross-dataset average AUC of 0.771 (vs. the teacher’s 0.796) with only 6.67M parameters. Furthermore, it supports real-time CPU inference (30+ FPS) and remains highly robust to JPEG compression and Gaussian blur, offering a practical solution for mobile forensic deployment.
: The detection of minority network attacks, such as User-to-Root (U2R), remains a challenge due to class imbalance. Traditional deep learning classifiers favor the majority class and exhibit large variations depending upon random initializations. We propose Proto-CBANet (Prototypical Class-Balanced Attention Network), which combines a 1D-CNN with an Average-Pooling Squeeze-and-Excitation block for feature recalibration, enabling distance-based reasoning in a latent metric space. To ensure statistical reliability, we evaluate our model across 60 random seeds. On NSL-KDD, Proto-CBANet achieves an average U2R recall of 61.07% ± 7.18%. To mitigate the trade-off between precision and recall for rare attacks, we further develop an advanced variant, HPN-GLF (Hyperspherical Prototypical Network with Global-Local Fusion). HPN-GLF integrates global learnable prototypes with hypersphere alignment, global-local semantic fusion, and dual-objective metric optimization. This advanced model reduces parameter count and FLOPs by 95% while significantly improving U2R precision from 12.95% to 59.76% and achieving the highest Macro F1 among compared methods. On NF-UNSW-NB15-v2, HPN-GLF attains a Macro F1 of 59.39% and high Shellcode recall (97.09%), indicating its promise for resource-constrained, imbalanced network security environments.
Chenxu Zhao· Proceedings of the 23rd Inte...· 0 citations
As deep learning generation technologies continue to evolve, Deepfake technology has become a major threat to information security, posing significant challenges to tampering detection. Existing detection methods for deep forgeries generally suffer from insufficient generalization capability and interpretability. To address these issues, this paper proposes a Deepfake tampering detection method based on a multimodal large model, termed LLM-FNP. This method integrates frequency-domain noise perception with visual-language understanding. First, the BayarConv module is employed to extract frequency-domain noise features from images, capturing anomalous noise patterns left by the forgery process, and a dual-branch encoder is used to process the original RGB images and frequency-domain feature maps separately. Second, a cross-attention mechanism is applied to achieve cross-modal fusion. Finally, the fused features are processed by a fine-tuned LLaVA large model, which outputs the detection result along with interpretable text. Experimental results on the DFDCP dataset show that the proposed method achieves an accuracy of 78.54% and an AUC of 81.50%. To evaluate the generalization capability of the method, cross-dataset testing on Celeb-DF and WildDeepfake yields accuracies of 73.5% and 71.8%, and AUCs of 76.4% and 75.1%, respectively, all outperforming classical methods such as ResNet+LSTM, further validating its strong generalization ability.
The rapid advancement of Generative Artificial Intelligence (GAI) has led to the proliferation of deepfake media, posing significant threats to digital security, privacy, and information integrity. To overcome this challenge, substantial research efforts have been directed toward developing automated detection techniques using deep learning methodologies. This study presents a comprehensive survey of deep learning-based deepfake detection methods, emphasizing lightweight and hybrid architectures designed for real-time deployment. The survey systematically categorizes the landscape of deepfake generation techniques and evaluates state-of-the-art detection frameworks, including: classical CNNs, efficient backbone architectures (MobileNet, EfficientNet), and spatiotemporal models (CNN LSTM/GRU). Furthermore, this study examines model compression techniques— including pruning and quantization — essential for resource-constrained deployment, and provides a structured analysis of benchmark datasets, major detection architecture categories, and persistent research gaps. By critically examining the trade-off between detection accuracy and computational latency, this paper identifies key open challenges and concludes by highlighting a research gap for probabilistically robust, lightweight frameworks, offering a roadmap for future research toward reliable, real-time deepfake forensics in unconstrained environments.
Azhar Abdulmughni, Nagi Al-shaibany· مجلة جامعة صنعاء للعلوم التط...· 0 citations