Kaya analyzes images, PDFs, and source code to detect tampering and AI-generated content.
python app.pyThe app will re-exec into .venv312 automatically if present.
.\.venv312\Scripts\python train_model.pyThis repo does not store training datasets. Download and place them locally as needed:
- Hugging Face: https://huggingface.co/datasets/basakdemirok/AIGCodeSet
- Back3474 GitHub dataset: https://github.com/Back3474/AI-Human-Generated-Program-Code-Dataset
After downloading, run train_model.py to populate datasets/training and retrain the model.
- Generated artifacts (datasets, uploads, reports, temp files, and model pickles) are ignored by git.
- Documentation lives at
/documentationin the running app.