P&ID PDF Processor with YOLO Detection

An intelligent PDF processing application that uses computer vision and OCR to automatically detect, classify, and extract text from Piping & Instrumentation Diagrams (P&ID). Built with a custom-trained YOLO model for shape detection and multiple OCR engines for text extraction.

🚀 Features

• Custom YOLO Model: Trained specifically for P&ID component detection
• Multi-OCR Support: Choose between PaddleOCR, EasyOCR, or custom OCR pipelines
• Interactive GUI: Drag-and-drop interface built with Tkinter
• Batch Processing: Process multiple PDF pages efficiently
• Export Functionality: Export results to Excel (XLSX) format
• Visual Detection: View detected components with bounding boxes and confidence scores
• Robust Error Handling: Multiple fallback strategies for reliable processing

📋 Requirements

System Requirements

• Python 3.8 or higher
• Windows/Linux/macOS
• Minimum 8GB RAM (16GB recommended for large PDFs)
• CUDA-compatible GPU (optional, for faster processing)

Dependencies

ultralytics>=8.0.0
opencv-python>=4.5.0
pandas>=1.3.0
matplotlib>=3.3.0
pdf2image>=2.1.0
pillow>=8.0.0
numpy>=1.21.0
tkinter (usually included with Python)
tkinterdnd2>=0.3.0
easyocr>=1.6.0
paddlepaddle>=2.4.0
paddlex>=2.1.0

Additional Requirements

• Poppler: Required for PDF processing
  • Windows: Download from poppler-windows
  • Ubuntu/Debian: sudo apt-get install poppler-utils
  • macOS: brew install poppler

🛠️ Installation

1. Clone the repository

git clone https://github.com/yourusername/pid-pdf-processor.git
cd pid-pdf-processor

2. Create virtual environment (recommended)

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Download and setup Poppler

• Update the poppler_path variable in PDFProcessor.py to match your installation

5. Place your trained YOLO model

• Put your best.pt model file in the models/ directory
• Or update the model path in get_model() function

📖 Usage

GUI Application

python main.py

1. Load PDF: Drag and drop a PDF file or use the browse button
2. Processing: The application will automatically detect P&ID components
3. Review Results: View detected shapes and extracted text in the data grid
4. Export: Save results to Excel format

Programmatic Usage

from PDFProcessor import get_data_from_pdf_easyocr

# Process PDF with EasyOCR
df = get_data_from_pdf_easyocr(
    pdf_path="your_pid_diagram.pdf",
    progress_callback=None,
    visualize='matplotlib'
)

# Export results
df.to_excel("results.xlsx", index=False)

🎯 YOLO Model Training

Dataset Preparation

1. Collect P&ID Images: Gather diverse P&ID diagrams
2. Annotation: Use tools like LabelImg or Roboflow
3. Classes: Define your P&ID component classes (e.g., valves, pumps, instruments, pipes)

Training Process

# Install Ultralytics
pip install ultralytics

# Train the model
yolo train data=pid_dataset.yaml model=yolov8n.pt epochs=100 imgsz=640

# Validate the model
yolo val model=runs/detect/train/weights/best.pt data=pid_dataset.yaml

# Run inference
yolo predict model=runs/detect/train/weights/best.pt source=test_images/

Dataset Structure

dataset/
├── images/
│   ├── train/
│   ├── val/
│   └── test/
├── labels/
│   ├── train/
│   ├── val/
│   └── test/
└── pid_dataset.yaml

Sample pid_dataset.yaml

path: ./dataset
train: images/train
val: images/val
test: images/test

nc: 8  # number of classes
names: ['valve', 'pump', 'instrument', 'pipe', 'tank', 'heat_exchanger', 'compressor', 'control_valve']

🔧 Configuration

OCR Engine Selection

Choose your preferred OCR engine by calling the appropriate function:

• EasyOCR (Recommended): get_data_from_pdf_easyocr()
• PaddleOCR: get_data_from_pdf()
• Memory-based: get_data_from_pdf_memory()

Visualization Options

# OpenCV visualization
get_data_from_pdf_easyocr(visualize='cv2')

# Matplotlib visualization (default)
get_data_from_pdf_easyocr(visualize='matplotlib')

# No visualization
get_data_from_pdf_easyocr(visualize=None)

📊 Output Format

The application generates a structured DataFrame with the following columns:

Column	Description
Shape	Detected P&ID component type
Label	Extracted text from OCR
X, Y	Top-left coordinates of bounding box
Width, Height	Dimensions of detected component
PDF Name	Source PDF filename

🐛 Troubleshooting

Common Issues

1. Model not found

   • Ensure best.pt is in the correct directory
   • Check file permissions

2. Poppler not found

   • Verify Poppler installation
   • Update poppler_path in configuration

3. OCR failures

   • Try different OCR engines
   • Check image quality and resolution
   • Ensure sufficient memory is available

4. Memory issues

   • Reduce batch size
   • Use sequential processing instead of parallel
   • Close other applications to free up RAM

Performance Optimization

• GPU Acceleration: Ensure CUDA is properly installed for faster inference
• Image Preprocessing: Adjust DPI and image enhancement parameters
• Model Optimization: Consider using YOLOv8s or YOLOv8m for better accuracy vs speed trade-offs

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Development Setup

# Install development dependencies
pip install -r requirements-dev.txt

# Run tests
python -m pytest tests/

# Code formatting
black .
isort .

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Ultralytics YOLOv8 for object detection
• EasyOCR for text recognition
• PaddleOCR for alternative OCR capabilities
• pdf2image for PDF processing

📞 Support

For questions, issues, or feature requests, please:

1. Check the Issues page
2. Create a new issue with detailed information
3. Contact: [email protected]

🚀 Roadmap

• Support for multi-page PDF processing
• Advanced P&ID component relationship mapping
• Integration with CAD software APIs
• Real-time processing capabilities
• Web-based interface option
• Docker containerization
• Cloud deployment options

---

Made with ❤️ for the Process Engineering Community