A privacy-preserving image forgery detection framework using Federated Learning. Combines FedYogi optimization with CNN and ResNet50 to handle non-IID forensic data. Trains models collaboratively without sharing raw images, improving stability and accuracy on CASIA 1.0.
python deep-learning numpy scikit-learn cnn pandas matplotlib transfer-learning resnet-50 federated-learning image-tampering-detection digital-image-forensics privacy-preserving-ai non-iid-client-heterogeneity fedyogi-optimization
-
Updated
May 26, 2026 - Jupyter Notebook