Zakaria Farih | Software Developer

GitHub Repo

About This Project

An AI-powered smart parking management system that leverages real-time computer vision, interactive 2D/3D visualization, and a scalable backend to optimize urban parking, reduce congestion and cut CO₂ emissions.

---

🚗 Core Functionality

• Real-Time Occupancy Detection
  A YOLOv11 model—trained on aerial images (drones, satellites)—processes live video frames to identify occupied vs. free spaces with high precision and low latency[^1].
• Interactive 2D/3D Visualization
  • 2D View: Polygons rendered via KonvaJS allow pan/zoom and spot-level interaction.
  • 3D Option: Modular Three.js integration (and Blender-modeled assets) for immersive layouts[^2].
• Backend & API
  Spring Boot REST/WebSocket services centralize data, enforce JWT security, and handle occupancy updates at scale[^3].

---

🎯 Objectives

1. Increase Parking Throughput
   Deliver real-time availability to drivers via camera feeds—no costly sensors or RFID required[^4].
2. Reduce Traffic & Emissions
   Minimize search time, easing urban congestion and lowering CO₂ output (aligned with UN SDG 11)[^5].
3. User-Friendly Management
   Intuitive GUI for spot definition (via QGIS/CAD→GeoJSON), reservations, and role-based administration.
4. Scalable, Modular Architecture
   Easily extendable for license-plate recognition, multi-level garages, cloud analytics, and edge-deployable inference.

---

🛠️ Technology Stack

• Vision & AI: Python, OpenCV, YOLOv11 (trained/augmented via Roboflow)[^6]
• Frontend: Vite + React TS, react-konva, Three.js, Material-UI
• Backend: Java Spring Boot, Spring Security (JWT), WebSocket, bucket4J rate limiting
• Data Layer: MySQL, Redis caching
• DevOps: Docker, Kubernetes (future), Railway CI/CD, ONNX export

---

🔍 Methodology & Results

1. Data & Model Training

• Collected/annotated aerial datasets (Roboflow)
• Trained YOLOv11l for a balanced speed/accuracy trade-off

2. Performance Metrics

• mAP: 97.6% Precision: 97.9% Recall: 94.8%[^7]
• Sustained 30–60 FPS inference with <1 s end-to-end latency

3. Usability & Reliability

• Pilot usability tests refined UI and import pipelines
• Penetration tests and GDPR-compliant data handling ensure security

4. Deployment

• Local demo with Spring Boot + Vite; live video feed integration
• Future edge deployment planned via ONNX/TensorRT

---

🌱 Sustainability & Impact

By transforming existing camera infrastructure into a smart parking network, this solution delivers an eco-friendly, cost-effective alternative to sensor-based systems—boosting mobility in both developed and developing urban contexts[^8].

---

For full technical details, refer to the project report (Sections: Resumen, Metodología, Desarrollo, Pruebas, Despliegue).

Footnotes :

[^1]: See training and evaluation details in Section 3 of the report.
[^2]: Visualization modules described in Section 4.
[^3]: API design and security in Section 5.
[^4]: Objective justification in Section 2.
[^5]: Environmental impact analysis in Section 6.
[^6]: Model augmentation process in Section 3.
[^7]: Performance benchmarks in Section 7.
[^8]: Sustainability discussion in Section 8.

Parking Management TFG

GitHub RepoDemo Video

Rubik's Cube Solver

Tech Stack: LibGDX, Kotlin, OpenCV, Python

This project allows users to input the state of a Rubik's Cube and receive step-by-step solving animations. It also features an interactive mode for scrambling and solving, and is open-source for community contributions.

GitHub RepoLive DemoDemo Video

Full-Stack Blogging Platform

Tech Stack: Spring Boot, Thymeleaf, JavaScript, HTML

A full-featured blogging platform that includes comments, reactions, user authentication, chat, and tagging. It is fully deployed and available to the public.

Worker Management System

Tech Stack: Elixir (Backend), React (Frontend)

Developed for Digital Data Real Time, this system handles worker tracking, expense management, work hour monitoring, and vehicle tracking. The project is under continuous development.

Elderly Home Management System

Tech Stack: React, Spring Boot, PostgreSQL

A personal project aimed at streamlining elderly care by managing patient data, staff assignments, and medical reminders. Currently a work in progress.

GitHub Repo

Pokémon Game & Pokédex Tracker

Tech Stack: Java, LibGDX

This project enables users to track, collect, and interact with Pokémon while building their own Pokédex.