Building a Real-Time License Plate Recognition System Using Big Data Technology
Bui Thanh Phuong - CH1902016
Smart cities are becoming increasingly prevalent in urban areas worldwide, particularly in the field of transportation. To effectively manage traffic-related tasks, it is essential to implement a license plate recognition system. Furthermore, with the exponential growth of big data, processing and analyzing this data require support from big data tools and artificial intelligence. Therefore, to utilize and optimize the available resources, this project aims to propose and implement a system using the RetinaNet model trained with the BigDL big data processing library. This system aims to identify license plates of moving vehicles in real-time using a Jetson Nano embedded computer connected to a camera.
Achievements:
- Proposed a smart traffic system using big data technology to recognize license plates of vehicles in real-time.
- Applied deep learning methods to increase the accuracy of license plate recognition on the big data platform.
- Utilized 2365 raw images with various sizes and qualities.
- Annotated 1750 images of vehicles in VOC format.
- Processed 30 video files ranging from 30 seconds to 1 minute in length, with resolutions from Full HD to 4K.
- Developed a browser extension to download vehicle images.
For more details, visit: https://fit.uit.edu.vn/index.php/tin-tuc/goc-hoc-tap/6484-xay-d-ng-h-th-...
Ha Bang - Media Collaborator, University of Information Technology
Translation: Nhat Hien
