Recipify

Recipify is a powerful receipt data extraction tool that combines YOLO11 for object detection and Tesseract OCR for text recognition. It automatically extracts key information from receipt images, such as vendor names, total amounts, items, and dates, to help you keep track of your expenses.

---

Features

• YOLO11 Object Detection: Detects receipt elements like shop names, total amounts, dates, and items.
• Tesseract OCR: Extracts textual data from the receipt images.
• Advanced Preprocessing: Enhances image quality for better OCR and object detection.
• Dataset Conversion: Converts annotated receipt datasets into YOLO format for training.
• CLI Support: Easy-to-use command-line interface for running the demo.

---

Setup and Installation

1. Clone the Repository

git clone https://github.com/subhashhhhhh/Recipify.git
cd Recipify

2. Set Up a Virtual Environment

Create and activate a Python virtual environment:

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

3. Install Recipify

Install the project and its dependencies:

pip install -e .

---

Usage

1. Run the Demo

To process a receipt image and extract details:

recipify-demo --image <path_to_receipt_image> --weights <path_to_yolo_weights>

Example:

recipify-demo --image example_receipt.jpg --weights yolo11n.pt

2. Train the YOLO11 Model

If you want to train YOLO11 on a custom dataset:

1. Ensure your dataset is structured as follows:

   dataset/
   ├── train/
   │   ├── images/
   │   └── labels/
   ├── val/
   │   ├── images/
   │   └── labels/
   

2. Update dataset.yaml:

   train: dataset/train/images
   val: dataset/val/images
   nc: 5
   names: ['shop', 'item', 'total', 'date_time', 'receipt']

3. Train the model:

   python -m ultralytics.yolo train --data dataset.yaml --weights yolo11n.pt --epochs 50

---

Folder Structure

Recipify/
├── recipify/
│   ├── __init__.py
│   ├── preprocessing.py       # Image preprocessing for OCR
│   ├── ocr.py                 # Text extraction using Tesseract
│   ├── classification.py      # Receipt type classification
│   ├── extraction.py          # Data parsing and extraction
│   ├── dataset_processing.py  # Convert annotations to YOLO format
│   ├── train.py
│   ├── dataset.yaml
├── dataset/                   # Custom dataset (images and labels)
├── requirements.txt           # Python dependencies
├── setup.py                   # Installation configuration
└── README.md                  # Project documentation

---

Model Evaluation

Confusion Matrix

Precision-Recall Curve

F1-Score Curve

Loss Curves

---

Development and Contribution

Testing

• Run tests with pytest:

  pytest

Contributions

Contributions are welcome! To contribute:

1. Fork the repository.
2. Create a new branch:

   git checkout -b feature-branch

3. Commit your changes:

   git commit -m "Add a new feature"

4. Push the branch:

   git push origin feature-branch

5. Open a pull request.

---

Dependencies

• Python 3.9+
• YOLO11: Ultralytics for object detection.
• Tesseract OCR: For text extraction.
• SpaCy: For named entity recognition (NER).
• OpenCV: For image preprocessing.

Install all dependencies via:

pip install -r requirements.txt

---

Acknowledgments

• Ultralytics: For the YOLO11 model.
• Google: For Tesseract OCR.
• OpenCV: For powerful image processing capabilities.
• Community: For contributing open datasets for receipts.

---

License

This project is licensed under the MIT License.