This is a program to extract information from Vietnamese ID card

1. Download model file follow link:

   https://hub.ultralytics.com/models/hgfIRTQBokYdGBQS7orm (Model 1 - use for extract text on front side of IDcard except portrait and QR code)

   Update new model: https://api.ultralytics.com/v1/predict/je3LTBqoLDRiZBtYRSYQ (Model 2 - use for extract QR code except text fields)

2. Dataset:

   • version 1 (not extract information from QR code): https://hub.ultralytics.com/datasets/EQ74fFtZdCei1GTLJRJF
   • version 2 (extract information from QR code): https://hub.ultralytics.com/datasets/G44KxW5Rce9ztGGqnI6X

3. Clone the module:

   git clone https://github.com/nguyen-tho/ID-card-extract-module.git

4. Install dependencies

pip install -r requirements.txt

5. Run process.py

   # run processing function in process.py
   python process.py
   #run REST API
   #python app.py
   # API is incomplete 
   #access http://127.0.0.1:5000 to observe the result

6. References

   • Label dataset tool: https://app.roboflow.com/
   • Training model tool: https://hub.ultralytics.com/
   • OCR extraction tool: https://github.com/pbcquoc/vietocr.git
   • QR code extract tool: https://pypi.org/project/qreader

7. New update

   • New model to detect QR code on Vietnamese ID card (depend on the detection model if they cannot detect QR code we cannot read issue date in the QR code)
   • Extract information from QR code such as issue date instead of create a new model to detect issue date at back-side of ID card
   • Update with REST API with Flask
   • Combine 2 models to improve the result
   • Update multi-processing with threads

8. Demo

   Input image

   

   Output image

   

   Output result (json)

   

   Confidence and extraction probability

9. Contact:

   Please contact with me at this email address to discuss: nguyencongtho116@gmail.com

10. New update (March 25th 2025) A simple UI using HTML, CSS and JS and an simple API using Flask

    # run app API
    python app.py

    To run app, we setup a localhost with port 5000

    localhost:5000
    #or
    127.0.0.1:5000