AI Pharaoh

AI Pharaoh is an integrated artificial intelligence platform that bridges modern machine learning with Ancient Egyptian cultural heritage. It combines computer vision, natural language processing, and augmented reality to support both hieroglyph interpretation and monument-based cultural exploration.

---

Overview

The system operates through two main pipelines:

Pipeline A — Hieroglyph Translation Raw inscription image → glyph detection (YOLOv8) → segmentation (SAM) → hierarchical classification (ConvNeXt, 1,080 Gardiner codes) → spatial ordering → transliteration → NLP translation → English output

Pipeline B — Monument Recognition & AR Monument photo → detection (YOLOv8m) → knowledge base lookup → AR 3D overlay with audio narration

---

Repository Structure

AI-Pharaoh/
├── src/                    # Core pipeline source code
│   ├── pharaonic_dashboard.py   # Streamlit UI
│   ├── config.py                # Configuration loader
│   └── process_images.py        # Headless batch processor
├── translation/            # NLP translation engine (beam search, SRL, discourse)
├── transliteration/        # Gardiner sign → phonetic conversion
├── orientation/            # Spatial ordering & reading direction
├── docs/                   # Full stage-by-stage documentation
├── mobile-app/             # Unity AR application source (C#)
├── mobile-backend/         # Colab-based API backend (FastAPI + ngrok)
├── desktop-launcher/       # Desktop application source
├── config.example.json     # Example configuration (copy to config.json)
├── requirements.txt        # Python dependencies
├── LICENSE                 # MIT License
└── .gitignore

---

Quick Start

# 1. Clone the repository
git clone https://github.com/your-username/AI-Pharaoh.git
cd AI-Pharaoh

# 2. Install dependencies
pip install -r requirements.txt

# 3. Copy and edit configuration
cp config.example.json config.json
# Update model paths in config.json to match your setup

# 4. Run the dashboard
streamlit run src/pharaonic_dashboard.py

Note: Pre-trained model weights (.pt/.pth) are not included in this repository. See Model Weights below.

---

Documentation

Full pipeline documentation is available in docs/:

Document	Description
Pipeline Overview	End-to-end system walkthrough
Stage 1: Input & Config	Image ingestion & config
Stage 2: YOLO Detection	Glyph proposal generation
Stage 3: SAM Segmentation	Mask refinement
Stage 4: NMS	Box + mask non-max suppression
Stage 5: Classification	ConvNeXt hierarchical classifier
Stage 6: Orientation	Reading direction estimation
Stage 7: Spatial Ordering	Reading sequence reconstruction
Stage 8: Transliteration	Gardiner → phonetic conversion
Stage 9: Translation	NLP translation engine
Stage 10: Monument Detection	Artifact recognition

---

Model Weights

The model weight files are not included in this repository due to size constraints. They are available on request. The following weights are needed:

Model	File	Size
YOLOv8 glyph detector	best.pt	~18 MB
SAM segmentation	sam_vit_b_01ec64.pth	~358 MB
Stage 1 classifier (29 families)	best_stage1_model.pt	~106 MB
Stage 2 classifier (1,080 codes)	stage2_best_model.pth	~110 MB
YOLOv8m monument detector	best.pt	~64 MB
Orientation classifier	orientation_model.pth	~9 MB

---

Key Results

Component	Metric	Value
Hieroglyph Detection (YOLOv8)	Test Accuracy	92.1%
Stage 1 Classification (29 families)	Test Accuracy	96.77%
Stage 2 Classification (1,080 codes)	Test Accuracy	93.2%
Combined End-to-End	Accuracy	89.44%
Monument Detection (YOLOv8m)	Test Accuracy	88.6%
Spatial Ordering	Sequence Accuracy	100% (synthetic)
Transliteration (known codes)	Phonetic Accuracy	98.2%

---

Built With

• Detection: Ultralytics YOLOv8
• Segmentation: Meta SAM ViT-B
• Classification: ConvNeXt Tiny (Facebook AI)
• NLP: Custom beam search + SRL pipeline
• AR: Unity ARFoundation + ARCore
• UI: Streamlit, React (desktop launcher)
• Backend: FastAPI, ngrok

---

License

This project is licensed under the MIT License — see LICENSE.

Citation

If you use this work in your research, please cite:

@misc{ai_pharaoh_2026,
  title={AI Pharaoh: An Integrated AI Platform for Ancient Egyptian
         Heritage Interpretation},
  author={AI Pharaoh Team},
  school={October University for Modern Sciences and Arts (MSA)},
  year={2026}
}