Nigerian License Plate Detection and Recognition System (NLPDRS)

An intelligent AI-powered system for detecting and recognizing Nigerian license plates using computer vision and deep learning. Built with a custom dataset from Nigerian roads, YOLOv8-nano for segmentation, and PaddleOCR for character extraction, the system is optimized for real-time deployment in law enforcement, intelligent surveillance, and road safety monitoring.

---

📌 Table of Contents

• Introduction
• Features
• System Architecture
• Dataset
• Model Pipeline
• API Usage
• Sample Use Cases
• Performance
• Challenges
• Recommendations and Future Work
• License

---

📖 Introduction

Nigeria’s growing road network faces increasing challenges including:

• Unregistered/stolen vehicles
• Manual vehicle checks
• Road crimes and traffic violations
• Lack of intelligent surveillance

Traditional methods are time-consuming and error-prone. The NLPDRS aims to modernize this process through AI-driven automated license plate detection and recognition, specifically tailored for Nigerian plate formats and environmental conditions.

---

✨ Features

• 🎯 Real-time License Plate Detection using YOLOv8
• 🔡 OCR-Based Character Extraction using PaddleOCR
• 🧠 Custom Trained on Nigerian Dataset
• 🕵️ Police Mode – Alerts for wanted vehicle plates
• 🛣️ FRSC Mode – Retrieves vehicle and owner details
• 📦 Web API powered by FastAPI for scalable integration
• 🔌 Ready for edge deployment on surveillance systems

---

🏗️ System Architecture

📷 Image Input │ ▼ 🎯 YOLOv8-nano (Plate Detection) │ ▼ 🔍 Cropped Plate Image │ ▼ 🔡 PaddleOCR (Character Recognition) │ ▼ 🧠 Decision Logic (Police / FRSC Mode) │ ▼ 📤 Output Result via FastAPI

---

🗃️ Dataset

• Collected by: EJAZTECH.AI
• Images: 300+ Nigerian vehicle images
• Sources: Various locations in Kano State, e.g., Sabon Gari, Farm Center, Kasuwar Dan Goro, etc.
• Resolution: Standardized to 640x640
• Annotation Tool: CVAT.AI
• Annotation Format: YOLO .txt
• Annotation Method: Segmentation-based (focus on 7-character license number)

---

🧠 Model Pipeline

• 📌 Detection Model: YOLOv8-nano

  • Trained specifically on Nigerian license plates
  • Handles distortion, complex lighting, varying angles

• 📌 OCR Engine: PaddleOCR

  • Extracts only the 7-character plate number
  • Works across both old and new plate formats

---

🔌 API Usage

POST /detect

Use this endpoint to submit an image and select your detection mode (Police or FRSC).

🧾 Form Fields:

Field	Type	Description
file	File	Vehicle image (JPEG/PNG)
mode	String	Either "Police" or "FRSC"
suspected_numbers	String	(Police Mode) Suspected plate numbers (newline-separated)
frsc_data	String	(FRSC Mode) Optional JSON: Plate → Vehicle & Owner Mapping

📂 Sample FRSC JSON:

{
  "ABC123KJ": {
    "owner": "John Doe",
    "vehicle": "Toyota Corolla, 2015, White"
  }
}

🧪 Sample Use Cases

🚓 Law Enforcement Mode (Police)

• Detects wanted or blacklisted plates
• Sends instant alerts to the user when a match is found

🛣️ FRSC Mode

• Retrieves owner and vehicle details from the provided database
• Assists in registration checks and ownership validation

🧾 Traffic Analytics

• Logs detected plates for movement tracking and traffic pattern analysis

🛰️ Edge Deployment

• Can be installed on highway cameras, drones, or checkpoints
• Enables autonomous, real-time monitoring

---

📊 Performance Metrics

Evaluation on held-out 20% test set:

Metric	Score
Precision	0.87
Recall	0.91
mAP@0.5	0.93
mAP@0.90	0.74

✅ High performance under diverse Nigerian conditions confirms model reliability and real-world readiness.

---

⚠️ Challenges Faced

• ❌ Bounding box annotations were inaccurate → switched to segmentation
• ❌ Standard OCR tools failed on Nigerian formats → adopted PaddleOCR
• 🚫 Difficulty integrating with CCTV APIs and hardware systems

---

🔭 Recommendations & Next Steps

• 📈 Expand dataset: More cities, states, lighting conditions, and plate types
• 🤝 Partner with FRSC & NPF: Access official APIs and registration data
• 🚀 Deploy on edge devices: Cameras, checkpoints, and mobile drones
• 📴 Offline version: Enable use in rural or internet-poor areas

---

👨‍💻 Developers

• 🧪 Project By: EJAZTECH.AI
• 👨‍🔬 Lead Developer: Ismail Ismail Tijjani
• 🏫 Institution: Bayero University, Kano
• 📆 Year: 2025

---

📜 License

This project is licensed under the MIT License — free to use, distribute, and build upon.

---

🤝 Contributing

We welcome collaboration and improvements!

• 🔱 Fork this repo
• 🔁 Create a Pull Request
• ✉️ Or reach out to our team directly

Your ideas, contributions, and feedback are truly appreciated!

---