Automated Vehicle Entry–Exit Analytics System

(Edge AI License Plate Recognition using NVIDIA Jetson Nano)

📖 Project Overview

This project implements an Automated Vehicle Entry–Exit Analytics System using edge AI on the NVIDIA Jetson Nano.

The system performs Automatic Number Plate Recognition (ANPR) to detect vehicles entering and exiting a facility by identifying license plates in real time.

Using deep learning–based object detection and OCR pipelines, the project demonstrates how embedded edge devices can perform real-time vehicle monitoring and access analytics without relying on cloud processing.

The project focuses on:

• Detecting vehicle license plates using deep learning object detection models
• Extracting license plate text using Optical Character Recognition (OCR)
• Running the full inference pipeline on the NVIDIA Jetson Nano edge device
• Evaluating performance trade-offs of different detection architectures on embedded hardware

---

⚙ System Capabilities

The system enables:

• Automated vehicle detection at entry and exit points
• License plate recognition using deep learning + OCR
• Edge inference using Jetson Nano (low latency)
• Vehicle entry–exit monitoring and analytics

Example applications include:

• Smart parking systems
• Automated gate access control
• Traffic monitoring
• Campus or facility vehicle logging

---

🧠 AI / Computer Vision Pipeline

The system pipeline follows these stages:

1. Image Capture

   • Camera input connected to the Jetson Nano

2. License Plate Detection

   • Deep learning models used for detection:
     • YOLO
     • SSD
     • Faster R-CNN

3. Plate Cropping

   • Detected plate region is isolated

4. Text Recognition

   • OCR techniques extract license plate numbers

5. Vehicle Entry–Exit Logging

   • Recognized plate numbers can be used for tracking and analytics

---

🛠 Technology Stack

• Hardware

  • NVIDIA Jetson Nano
  • Camera module

• Software

  • Python
  • OpenCV
  • PyTorch / TensorFlow
  • YOLO object detection
  • OCR pipeline

• Deployment

  • Edge AI inference on embedded hardware

---

📂 Folder Structure

A simplified view of the repository:

.
├── Assembly/                     # LaTeX report build files
├── Code/                         # Source code
│   └── LicencePlateDetection/    # ANPR / LPR experiments & apps
├── JetsonNano/                   # Jetson Nano-specific code & setup
│   ├── Code/                     # Scripts (Arduino/Jetson/etc.)
│   ├── Images/                   # Sample images & figures
│   ├── System/                   # Configuration & system setup
│   └── Contents/                 # Supporting files
├── Manual/                       # Project manual (LaTeX chapters)
├── MLbib/                        # Literature, manuals, and references
├── Poster/                       # TikZ poster files
├── Presentations/                # Presentation slides
│   └── Literatures/              # Literature review & references
│       ├── LiteratureReview.pdf
│       └── References/           # Collection of research PDFs
├── ProjectManagement/            # Checklists, evaluation docs
├── report/                       # Final written report
└── README.md                     # This file

📚 Literature

The literature collection under Presentations/Literatures/References/ covers:

• Deep learning approaches for license plate detection
• YOLO, SSD, and Faster R-CNN comparisons
• OCR-based text extraction methods
• Jetson Nano and edge AI deployment studies

---

👥 Contributors

• Gautam Ramesh (Matriculation Number: 7026787)
• Shubhankar Dumka (Matriculation Number: 7027187)
• Siddhanth Sharma (Matriculation Number: 7027189)

---

📌 Notes

• This project is developed for academic purposes under the Master’s program in Business Intelligence and Data Analytics.
• Future work will include benchmarking model accuracy and speed, and improving OCR reliability on real-world data.