A complete pipeline for extracting and processing information from receipt images using Optical Character Recognition (OCR). This project leverages cutting-edge technologies such as PaddleOCR, VietOCR, and PICK for text detection, recognition, and structured information extraction. A web interface is provided for seamless user interaction.
The project workflow consists of the following steps:
- Text Detection: Accurate detection of text regions using PaddleOCR.
- Rotation Correction: Automatic alignment of rotated text using OpenCV and MobileNetv3.
- Text Recognition: Recognize text content with VietOCR.
- Information Extraction: Extract structured data using PICK.
- Clone the repository:
git clone https://github.com/johnPa02/ocr_receipts.git
cd ocr_receipts
- Install dependencies:
pip install -r requirements.txt
Start the server:
python app.py