License Plate Detection and Recognition

Aritra Roy Gosthipaty

Aritra Roy Gosthipaty

Kolkata, West Bengal

A full stack data science project which helps in the localisation of a license plate and extracting the number of the plate when provided with a vehicle image. The models are exposed in REST API and the API endpoint is consumed by a Mobile Application. ...learn more

Project status: Under Development

Mobile, Artificial Intelligence

Intel Technologies
Intel Python

Code Samples [1]Links [2]

Overview / Usage

A full stack Data Science project. This project has two phases, first being an object detector which detects the location of a license plate provided an image of a vehicle, second is an optical character recogniser which extracts the license number from the license plate. The two models work together. We need to input an image of a vehicle, the object detector detects the license plate and then sends the cropped license plate image to the optical character recogniser. The models are exposed as REST API with the help of Flask. The API endpoint is consumed by an Android Mobile Application. In production, this application can come in handy in the security sectors. The project serves as a black box to people which extracts the license number from a vehicle image. The license number in string format can be used efficiently and to ones purpose.

Methodology / Approach

Object Detector: The model has been trained with the help of Tensorflow Object Detection API. The model that has been trained uses the hybrid architecture of Single Shot Detection and Mobile Net.

Optical Character Recogniser: Have used a simple Support Vector Machine for the digit classification. The segmentation of each digit out from a vehicle plate is done by some basic computer vision tweaks.

**REST API: **The models have been exposed as REST API made with the help of Flask. The API endpoints help the world to use the models as a black box in their own way of implementation.

Mobile Application: The API is consumed by a mobile application that serves as the client. The application can click pictures and send it to the necessary API endpoint and get the license number or the license plate in return.

Technologies Used

Object Detection:

  • Intel Python
  • TensorFlow Object Detection API
  • Colab

Optical Character Recogniser:

  • Intel Python
  • scikit-learn

REST API:

  • Flask

Mobile Application:

  • Flutter

Repository

https://github.com/ariG23498/license-plate-detection

Collaborators

There are no people to show.

Comments (0)