A privacy-first browser extension that detects deepfake videos and audio in real-time using a local efficient deep learning model.
- Local Inference: Uses ONNX Runtime Web to run AI models directly in your browser. No data leaves your machine.
- Multimodal Detection: Analyzes both Video (Visual Artifacts) and Audio (Spectral Consistency).
- Modern UI: Glassmorphic overlay seamlessly integrates with YouTube, Twitter, and other video platforms.
- Optimized: Quantized models ensure low latency (<100ms) and minimal battery usage.
- Python 3.8+
- Node.js 16+
-
Setup Environment:
./scripts/setup.sh
-
Download Datasets (Required for training):
./scripts/download_datasets.sh
-
Train Model:
./scripts/train_all.sh
-
Build Extension:
./scripts/build_extension.sh
-
Load in Browser:
- Open Chrome/Edge and go to
chrome://extensions. - Enable "Developer Mode".
- Click "Load unpacked" and select the
extension/distfolder.
- Open Chrome/Edge and go to
- ML Backend: PyTorch (EfficientNet + ResNet + LSTM)
- Inference: ONNX Runtime Web (WASM)
- Frontend: Vanilla JS + Glassmorphism CSS
MIT