vntanh1406/Egypt-License-Plate-Classification

# Egyptian License Plate Classification System ## 1. Project Description This project develops an **automated classification system for Egyptian vehicle license plates** based on their **color-coded background**, a standard feature in Egypt's vehicle registration system. The model classifies license plates into **six official categories**: - **Private** (Blue) - **Taxi** (Orange) - **Commercial** (Red) - **Public Transport** (Grey) - **Diplomatic** (Green) - **Tourist/Temporary** (Yellow) The system uses **cropped license plate images** extracted from real-world Egyptian traffic videos. The classification is formulated as a **6-class supervised learning task** using deep learning, with **color** being the primary discriminative feature. Three models were evaluated using **5-fold stratified cross-validation**: - **Simple CNN** (custom lightweight architecture) - **EfficientNet-B0** (pretrained, transfer learning) - **MobileNetV2** (pretrained, lightweight baseline) **Simple CNN achieved the best performance** with near-perfect accuracy (0.99 average, 1.00 best fold) and fastest training time. > **Author**: Vo Ngoc Tram Anh > **Date**: October 27, 2025 --- ## 2. Dataset Info ### 2.1. Data Sources The dataset was manually extracted from **24 real-world YouTube videos** of Egyptian traffic, covering diverse conditions (day/night, urban/tourist areas, weather, lighting). Videos were selected to ensure natural class distribution and environmental variability. **Video List**: 1. youtube.com 2. youtube.com 3. youtube.com 4. youtube.com 5. youtube.com 6. youtube.com 7. youtube.com 8. youtube.com 9. youtube.com 10. youtube.com 11. youtube.c …