Traffic Sign Detection AI

Overview

This repository hosts the code and resources for an advanced AI-driven system developed for the detection and recognition of traffic signs. Our model leverages the YOLOv8 neural network architecture and is trained on a robust dataset from Mapillary, supplemented with local Hong Kong traffic sign images.

Objectives

• Enhance the accuracy of traffic sign detection and recognition, ensuring real-time operation.
• Equip the system to handle diverse environmental conditions, with a focus on Hong Kong's unique traffic regulations and infrastructure.

Methodology

1. YOLOv8 Model: Employing the YOLOv8 model for efficient and accurate real-time object detection.
2. Dataset: Utilizing a comprehensive dataset from Mapillary, enriched with local Hong Kong traffic sign images.
3. Training: Conducting intensive training using an NVIDIA Geforce RTX 4080 graphics card.
4. Deployment: Implementing the model in a user-friendly web interface for real-time traffic sign detection.

Findings & Results

• The system showed a high success rate in detecting standard traffic signs under typical conditions.
• Specific challenges were identified in recognizing unique Hong Kong-specific traffic signs and electronic displays.

Team Members

• Tony (hospotho)
• Soman Fung (skytim12)
• Flucus ~ NightSky (Flucus)
• Chan Ka Wing (KelvinSloth)

References

• The training data can be accessed here.

License

This project is licensed under the terms of the Apache.