GCME AI-Recognition Network for Engineering Technology
Precision in Every Connection

GARNET is an AI-powered tool designed to automate symbol detection, classification, and connectivity analysis in Piping and Instrumentation Diagrams (P&IDs). Built for engineers and maintenance teams, it combines state-of-the-art object detection (YOLOv11/YOLOv8) with graph-based analytics to transform P&ID workflows.

• Symbol Detection: Identify valves (gate, globe, check), pumps, tanks, and more using YOLOv11/YOLOv8
• SAHI Integration: Slicing Aided Hyper Inference for accurate detection on large images
• Text Recognition (OCR): Extract text annotations from symbols using multiple OCR engines
• Graph Construction: Build connectivity graphs from detected symbols and pipes
• Interactive Review: Human-in-the-loop validation and correction of results
• Export Options: Results available in JSON, YOLO, COCO, LabelMe, or PDF formats

• Interactive canvas with pan/zoom and minimap support
• Object detection visualization with color-coded bounding boxes
• Object editing (create, update, delete) directly on canvas
• Review workflow with accept/reject and visual status indicators
• Batch processing with queue management and progress tracking
• Undo/redo support for all editing operations
• Keyboard shortcuts for efficient navigation
• Dark/light theme toggle
• Confidence filtering and class visibility controls

• Automated counting and symbol type classification
• Model caching for faster subsequent detections
• Results caching with TTL for improved performance
• Automatic cleanup of temporary files
• Health monitoring endpoint
• Environment-configurable for development and production

• Python 3.9+
• Node.js 18+ (for frontend)
• Bun or npm (package manager)
• Git

1. Clone the Repository

   git clone https://github.com/may3rd/GARNET.git
   cd GARNET

2. Install Dependencies

   # Backend
   cd backend
   pip install -r requirements.txt
   cd ..
   
   # Frontend
   cd frontend
   bun install  # or: npm install
   cd ..

3. Configure Environment

   # Backend
   cp .env.example .env
   # Edit .env as needed
   
   # Frontend
   cd frontend
   cp .env.example .env.local
   # Edit .env.local as needed
   cd ..

4. Start the Application

   # Terminal 1 - Start API backend
   cd backend
   uvicorn api:app --reload --port 8001
   
   # Terminal 2 - Start React frontend
   cd frontend
   bun run dev  # or: npm run dev

This is the primary mode for interactive P&ID analysis. The React frontend provides a modern UI for uploading images, configuring detection parameters, reviewing results, and exporting data.

• Detection Mode: Symbol detection and OCR with interactive review
• Pipeline Mode: End-to-end P&ID digitization including connectivity analysis

Run inference on multiple P&IDs using garnet/predict_images.py:

python garnet/predict_images.py \
    --image_path path/to/pids_folder \
    --model_type yolov8 \
    --model_path path/to/model_weights.pt \
    --output_path results/

Comprehensive P&ID analysis pipeline available in garnet/pid_extractor.py:

python garnet/pid_extractor.py \
    --image path/to/pid_image.png \
    --out output/ \
    --stop-after 11

Detailed documentation is available in the docs/ directory:

• Frontend Structure
• Backend Structure
• API Endpoints
• Data Flow
• Pipeline Stages
• Environment Variables
• Model Training
• Dataset Format

Additional planned outcomes from the GARNET project include:

• MTO for Valves: Automated generation of material take-off lists for all detected valve types
• Line List: Extraction and tabulation of pipeline data including line tags, sizes, service, and connected equipment

Contributions are welcome! Please fork the repository and submit a pull request. For major changes, open an issue first to discuss your ideas.

This project is licensed under the MIT License. See LICENSE for details.

For questions or collaborations, contact:

• Maetee Lorprajuksiri - may3rd@gmail.com
• GCME (GC Maintenance and Engineering Co., Ltd.) - www.gcme.com