Fetching metadata from the HF Docker repository...
-
feat: initial project setup with audio/video detection modules and fusion
Aryanaideveloper +1: Claude Opus 4.6 - Initialize project structure for Multimodal Deepfake Detection
- Add Nes2Net_ASVspoof_ITW for audio deepfake detection (SOTA EER 1.49%)
- Add GenConViT for video deepfake detection (98.5% accuracy on DFDC)
- Create audio/video inference wrappers with normalized scoring
- Implement late fusion module with multiple strategies (weighted avg, voting, etc.)
- Add research analysis, implementation plan, and task tracker docs
- Add setup scripts for conda environments and weight downloads
- Add .gitignore for data, weights, and build artifacts
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>