buildborderless/OpenSight-Deepfake-Detection-Models-Playground at 0e2cdc4906eb4f67ab48ec61749fbaabacd8574d

OpenSight-Deepfake-Detection-Models-Playground

Ctrl+K

5 contributors

History: 254 commits

LPX55

feat: dynamically initialize model weights using MODEL_REGISTRY and adjust weights for specific models

0e2cdc4 about 2 months ago

agents
feat: dynamically initialize model weights using MODEL_REGISTRY and adjust weights for specific models about 2 months ago
forensics
docs: expand README with roadmap, feature status, and AI content detection tools; update requirements for transformers about 2 months ago
hf_inference_logs
refactor: rename image preprocessing function for clarity and update HTML output for consensus display about 2 months ago
models
refactor: reorganize agent structure by moving models to agents directory, update logging level, and enhance .gitignore for model files about 2 months ago
utils
feat: enhance model registration by adding architecture and dataset metadata to the ModelEntry class about 2 months ago
.gitattributes

1.52 kB

initial commit 11 months ago
.gitignore

184 Bytes

chore: update .gitignore to include additional Python bytecode files and inference log directories about 2 months ago
README.md

27.7 kB

feat: implement new app structure with enhanced ensemble prediction capabilities and integrate multiple models for image analysis about 2 months ago
app.backup.py

19.6 kB

refactor: update import paths for forensics utilities and enhance logging functionality in inference data about 2 months ago
app.py

24.6 kB

major(feat): implement streaming ensemble prediction to enhance real-time model inference and update interface for live results about 2 months ago
app_mcp.old.py

30 kB

feat: implement new app structure with enhanced ensemble prediction capabilities and integrate multiple models for image analysis about 2 months ago
requirements.txt

565 Bytes

chore: update requirements to remove index-url and ensure compatibility with GitHub source for transformers about 2 months ago