License-plate-recognition

Overview

This project uses a model trained with rear license plates from Ontario, Canada, and ultimately achieves object detection in an Android interface. The goal of the project is to utilize object detection technology to create a commercial model for automating parking lot entry/exit and checkout processes. This project employs Python, Kotlin, Node.js, and MongoDB.

• License Plate Detection

• Search and Checkout Interface

Workflow

Get Started

The YOLO model was trained with Python for object detection, and the trained model is loaded onto an Android interface to detect license plates and license plate numbers. The detected license plate images and numbers are processed via Node.js and added to MongoDB.

1. Data Collection and Preprocessing:

   • Data was collected using a Canon R6 and an iPhone XR (all photos taken in landscape orientation). The images were adjusted to 640x640 pixels using Lightroom, and quality was reduced to accelerate model training.

2. Data Labeling(CVAT, Label Studio....):

   • Export to YOLO format

3. Model Training uses Python (YOLOv11)

4. MongoDB Create & Setting

5. Backend Setting (.env)

   DB_USER=AAAA
   DB_PASSWORD=BBBB
   DB_NAME=CCCC
   DB_USE=ProjectName
   DB_DEFAULT_PORT=3005
   

6. Android Setting

• gradle.properties

  Backend_URL=Backend_URL = http://YourIPAddress:Port/api/
  

• Constants

  const val MODEL_PATH = "carDetectModel/YourYoloModel.tflite"
  const val LABELS_PATH = "carDetectModel/custom_label.txt"
  

API Endpoints

Check more API detail at ./API_DOCS