A Lightweight and Interpretable Deepfakes Detection Framework
Muhammad Umar Farooq, Ali Javed, Khalid Mahmood Malik, Muhammad Anas Raza
TL;DR
The paper tackles the lack of a unified, lightweight detector for all major deepfake types by fusing novel heart-rate features with robust facial landmark cues and classifying with an interpretable XGBoost model. It demonstrates that this hybrid feature approach achieves strong detection on WLDR, approaching the performance of deeper networks while maintaining transparency and efficiency. The main contributions are the novel heart-rate features, their fusion with landmark descriptors, and the demonstration that a gradient-boosted tree ensemble can generalize well across face-swap, lip-sync, and puppet-master deepfakes. This work offers a practical, scalable solution for real-world deepfake screening with potential for cross-dataset extension and deployment in settings requiring quick, interpretable decisions.
Abstract
The recent realistic creation and dissemination of so-called deepfakes poses a serious threat to social life, civil rest, and law. Celebrity defaming, election manipulation, and deepfakes as evidence in court of law are few potential consequences of deepfakes. The availability of open source trained models based on modern frameworks such as PyTorch or TensorFlow, video manipulations Apps such as FaceApp and REFACE, and economical computing infrastructure has easen the creation of deepfakes. Most of the existing detectors focus on detecting either face-swap, lip-sync, or puppet master deepfakes, but a unified framework to detect all three types of deepfakes is hardly explored. This paper presents a unified framework that exploits the power of proposed feature fusion of hybrid facial landmarks and our novel heart rate features for detection of all types of deepfakes. We propose novel heart rate features and fused them with the facial landmark features to better extract the facial artifacts of fake videos and natural variations available in the original videos. We used these features to train a light-weight XGBoost to classify between the deepfake and bonafide videos. We evaluated the performance of our framework on the world leaders dataset (WLDR) that contains all types of deepfakes. Experimental results illustrate that the proposed framework offers superior detection performance over the comparative deepfakes detection methods. Performance comparison of our framework against the LSTM-FCN, a candidate of deep learning model, shows that proposed model achieves similar results, however, it is more interpretable.