Hello, I'm
PhD Researcher — Traffic Sign Detection & Text Recognition in Arabic-Latin
I am a PhD researcher at the Faculty of Applied Sciences, Ait Melloul, specializing in Embedded Systems and Digital Services. My doctoral research focuses on Traffic Sign Detection and Text Recognition in Arabic-Latin scripts using deep learning techniques — addressing the unique challenges of multilingual road signage in North Africa and the Arab world. With over 1 year of professional experience, I combine my expertise in Artificial Intelligence, Computer Vision, IoT, and Full-Stack development to build innovative real-world solutions. From smart agriculture systems with biometric attendance to IoT-connected infrastructure monitoring, I bridge the gap between cutting-edge research and practical engineering.
My PhD research investigates advanced deep learning architectures for real-time detection and recognition of traffic signs containing Arabic and Latin text. This bilingual OCR challenge requires specialized models capable of handling mixed-script environments, varying fonts, weathered signs, and complex backgrounds typical of Moroccan and North African road infrastructure. The work combines state-of-the-art object detection (YOLO, Faster R-CNN) with transformer-based text recognition models to achieve robust multilingual sign understanding for autonomous driving and intelligent transportation systems.
Faculty of Applied Sciences, Ait Melloul
Faculty of Applied Sciences, Ait Melloul
Faculty of Applied Sciences, Ait Melloul
Faculty of Applied Sciences, Ait Melloul
Lycée Qualifiant Sidi Moussa
Web & IoT Developer
Web Developer
Mobile Developer – Smart Attendance App
AgriWise is a revolutionary Android application developed to modernize the attendance system in the agricultural sector. This comprehensive solution combines the latest biometric technologies to offer precise and secure work time management. The app supports triple authentication methods — fingerprint scanning, NFC card reading, and AI-powered facial recognition — making it adaptable to various work environments, even in remote agricultural fields with limited connectivity.
A complete digital solution developed for Risouss Agricole, a leading company in the production and export of high-quality fruits and vegetables. The project encompasses a modern corporate website showcasing the company's 12+ years of expertise across 113 hectares of production, managing 7,500 tons annually. Additionally, an internal management platform was built to streamline personnel management, inventory tracking, and operational workflows.
An end-to-end IoT solution developed for FromTelecom to enable real-time monitoring of connected water tanks and GPS tracking for vehicle fleets. The system uses ESP32 microcontrollers and various sensors to collect data on water levels, temperature, and tank conditions, transmitting information to a centralized web dashboard. The GPS tracking module provides live vehicle location, route history, and geofencing alerts for fleet management.
A cutting-edge deep learning research project addressing the challenge of detecting traffic signs and recognizing bilingual text (Arabic and Latin) in real-world road environments. The system implements YOLO-based object detection for real-time sign localization combined with transformer-based OCR models specifically trained for mixed Arabic-Latin script recognition. Custom datasets were collected from Moroccan road infrastructure, featuring weathered signs, varying fonts, and complex backgrounds. The model achieves robust performance across different lighting conditions, occlusion levels, and sign degradation states, contributing to the advancement of intelligent transportation systems in multilingual regions.
Development of a comprehensive IoT system for intelligent agricultural farm management. The system integrates multiple sensors (soil moisture, temperature, humidity, light) with ESP32 microcontrollers to monitor farm conditions in real-time. Data is transmitted via MQTT protocol to a central server, providing farmers with actionable insights through a web-based dashboard for irrigation scheduling, climate control, and crop health monitoring.
A comparative study analyzing trajectories collected from LIDAR sensors and RGB cameras with parallel processing capabilities. The project implements simultaneous data acquisition from both sensor types, applies trajectory extraction algorithms, and performs quantitative comparison of accuracy, precision, and computational cost. Parallel processing techniques are used to handle the large volumes of point cloud and image data efficiently.
A full-stack web application for managing gym facilities and training centers. The platform handles member registration, subscription management, trainer scheduling, attendance tracking, and financial reporting. Features include a responsive dashboard for administrators, member portal for booking sessions, and automated notification system for subscription renewals and class reminders.