Can a Parking System Be Smarter Than Humans? Designing a Secure IoTBased System with QR Code and YOLO
Description
This dataset accompanies the study “Can a Parking System Be Smarter Than Humans? Designing a Secure IoT-Based System with QR Code and YOLO”, conducted by Tamaris Roulina Silitonga from Politeknik Negeri Batam, Indonesia. The research focuses on developing and testing a secure, intelligent parking system that integrates YOLOv8 for license plate recognition, Optical Character Recognition (OCR), QR code authentication, and IoT-based automated gate control into a unified architecture. The system was designed to address limitations in traditional parking management—such as manual data entry, human error, and low operational efficiency—by combining deep learning and IoT technologies for automation, reliability, and security. The dataset and documentation include performance results, architecture diagrams, and implementation details of the developed modules: License Plate Detection Module using YOLOv8 and PaddleOCR. QR Code Authentication Module employing GM66 QR scanner as a fallback mechanism. IoT-Based Gate Control using ESP32 microcontroller and RESTful API for automated barrier operation. Laravel-Based Web Dashboard for real-time monitoring, data management, and reporting. Experimental evaluations were conducted in a simulated parking environment, achieving: 100% success rate in automated barrier control. 9–10 seconds average response time for license plate recognition. 3–5 seconds response time for QR code validation. These results demonstrate the system’s reliability and potential scalability for real-world deployment in smart city infrastructures. The dataset can be used as a foundation for further research in IoT automation, computer vision, and intelligent transportation systems, and may assist researchers or developers aiming to enhance smart parking, security automation, or urban mobility solutions.
Files
Steps to reproduce
Prepare the Development Environment Install Python 3.11, Google Colab, and required libraries: ultralytics, paddleocr, opencv-python, and requests. Set up a Laravel 10 web application with MySQL 8 as the database and configure the .env file for local or VPS hosting. Install the ESP32 board support in the Arduino IDE and required libraries for HTTP communication. YOLOv8 Model Training Collect and label at least 300 Indonesian vehicle license plate images under various lighting and angle conditions using LabelImg. Upload the dataset to Google Colab and train YOLOv8 with: Image size: 416×416 Epochs: 50 Batch size: 8 Early stopping: 10 Export the trained weights (best.pt) for integration into the detection module. System Integration Connect a webcam to capture vehicle images. Use YOLOv8 + PaddleOCR for plate detection and text extraction. Integrate the GM66 QR scanner to handle fallback authentication. Configure the ESP32 to send HTTP requests to the Laravel server for gate control. Web Application Setup Deploy the Laravel backend on a VPS or local server (e.g., alpu.web.id or localhost). Implement RESTful APIs for communication between the ESP32 and the web dashboard. Add modules for user management, vehicle registration, QR generation, and real-time monitoring. Testing and Evaluation Simulate vehicle entry and exit using registered and unregistered license plates. Measure detection accuracy, QR scan success rate, and system response time using a timer or log file. Compare both authentication methods under various conditions to validate performance. Expected Results License plate recognition: 9–10 seconds average response time, 99%+ detection accuracy. QR authentication: 3–5 seconds response time, 100% success rate. Gate control: 100% success via ESP32 communication.
Institutions
- Politeknik Batam