Wissemtriki/Tunisian-License-Plate-Recognition-AI

AI-powered Tunisian License Plate Recognition system developed during my internship at Horizop Energy. It performs plate detection, segmentation, extraction, and OCR using a custom collected dataset, with preprocessing and deep learning models for accurate recognition. 🇹🇳🔍 Tunisian License Plate Recognition – Horizop Energy Deep Learning system for detecting, extracting, and recognizing Tunisian vehicle license plates. Developed during my engineering internship at Horizop Energy. 📌 Project Overview This project implements a complete end-to-end License Plate Recognition (LPR) pipeline exclusively for Tunisian license plates. It includes: Plate detection using YOLO-based models Plate extraction & segmentation Character recognition (OCR) using a trained deep-learning model (ocrmodel.h5) Data preprocessing, cleaning, and analysis through Jupyter notebooks Character classification dataset for training OCR models All stages of the system are documented and implemented inside organized Jupyter notebooks. 📁 Repository Structure 📦 Horizop-IA-License-Plate-Recognition ┣ 📁 Notebooks/ ┃ ┣ Data Preprocessing & Cleaning - Horizop_version.ipynb ┃ ┣ Licence Plate Detection and Extraction - Horizop_version.ipynb ┃ ┣ Licence Plate Recognition - Horizop_version.ipynb ┃ ┣ Modeling for Licence Plate Recognition - Horizop_version.ipynb ┃ ┗ Main Script - Horizop_version.ipynb ┣ 📁 LP_extraction_test/ ┃ ┗ Sample test images used for plate extraction ┣ 📁 Characters-Classification-Data/ ┃ ┣ train/ ┃ ┗ val/ ┣ 📄 ocrmodel.h5 — Trained OCR model ┣ 📄 darknet-yolov3.cfg — Detection model configuration ┣ 📄 classes.names — YOLO classes for Tunisian plates ┗ 📄 README.md ⭐ Key Features 🔍 License Plate Detection using YOLO (V3 architecture) ✂️ Plate Extraction & Segmentation 🔡 Deep Learning OCR with custom-trained character classifier 🧹 Dataset cleaning, augmentation, and preprocessing 📊 Multiple structured notebooks for transparency and reproducibility 🖼️ Real test images included for validation 🧠 Technologies Used Python, OpenCV TensorFlow / Keras YOLOv3 (darknet-style config) NumPy, Pandas Scikit-learn Matplotlib & Seaborn Jupyter Notebook 🧪 Notebooks Explained 1️⃣ Data Preprocessing & Cleaning Includes dataset filtering, augmentation, no …