Student ID Card Recognition is my AI's project for scanning and extracting information from DUT's Student ID Card with Image Processing and state-of-the-art Deep Learning's model (YOLO + Transformer).
Student ID Card Recognition can be applied with License Plate Recognition to build a parking lot for Da Nang University of Technology.
- Image Processing with OpenCV
- Object Detection with YOLOv4-tiny
- Text Recognition with Transformer
- Flask-Python to create API and build website
- Docker for deployment
Current features:
- The main function of system is to identify and extract information from Student ID Card photo. Users can upload a card
image to the system, and this system will return extracted information from the photo, for example:
- ID Number
- Full name
- Class name
- Date of birth
- Faculty
- Year
To-do list:
- Use an Object Detection model (YOLO, SSD, etc) for scanning ID Card in the input image instead of using Image Processing.
- Add one more function to this system: Vietnamese ID Card Recognition
Because I don't have a server with GPU, so the processing time is quite long. The average processing time is:
- Server without GPU(4 Cores, 4 GB RAM): 2 - 5 seconds/photo.
- The input image must have 4 clear angles and its background should be white or gray and not contain anything else :disappointed_relieved: to make sure the ID Card can be seen.
- All information fields must be visible, readable, unmodified and not blurred.
- The input image size does not exceed 5 MB, and the minimum resolution is approximately 640x480 to ensure the confident rate.
- The ratio of Student ID Card area must be at least 2⁄3 of the total image area.
This website is temporarily unavailable because my Azure account has no credit to keeping this service 😅
I have built a simple website to show how it works. However, to use this website you need a Student ID Card as your input image. So, please download the test-img folder then use one of them. Sorry for this inconvenience 😞
Project is no longer continue because I'm dying with my mind 😢
Created by @dvt - feel free to contact me!