[CVPR 2020] A Large-Scale Dataset for Real-World Face Forgery Detection
Deepfakes Video classification via CNN, LSTM, C3D and triplets [IWBF'20]
Model and application for deepfake detection using a hybrid approach (spatial + frequency-based)
Production-grade deepfake detection — 3-model ensemble (Xception + EfficientNet-B4 + ViT), calibrated confidence, forensics suite, REST API, and video AV-sync/identity tracking.
Composites Progressive GAN-generated face patches into real images using homographic alignment and Poisson blending. Donor/recipient framework for forensics and data augmentation research.
Deepfake Detection Using Deep Learning
DeepFake detection with Deep Tree Network (DTN) architecture - 94.5% accuracy, 45 FPS real-time processing with zero-shot learning
Thesis topic of M.Tech final year from IIIT Allahabad
Official PyTorch implementation of "Depthtective: A Depth-Aware Framework for Spatio-Temporal Deepfake Detection" (Array, 2026). Detects manipulated videos from the RGB and depth residuals of just two adjacent frames.
Refining Forgery-Aware Prompts for Deepfake Detection with Pattern Blended Samples
To associate your repository with the face-forensics topic, visit your repo's landing page and select "manage topics."