home / projects / DeepShield: Deepfake Detection

DeepShield: Deepfake Detection_

Robotics & vision · Jul – Nov 2023 · paper · by Nithin Jambula

A CNN video model that detects face-swap deepfakes, with 22% better accuracy and 18% fewer false positives. A synthetic video generator added 40% more variety to the training data. The research, covering deepfake and counterfeit currency detection, was published at IEEE iSAI-NLP 2025. cite →

Built with

PyTorch · CNN · GAN · OpenCV

More robotics & vision