yahia997/PavementEye

A sustainable, cost-effective solution for automating pavement assessment in Alexandria, Egypt. Using computer vision (YOLO v8), cloud computing, and OpenStreetMap integration, it provides a real-time dashboard to visualize crack severity and intelligently prioritize road repairs. --- ## 🏆 Project Context > **Huawei Developer Competition Northern Africa 2025** **PavementEye** is an end-to-end streaming data pipeline that leverages Deep Learning and Big Data technologies to automate road inspection. It was deployed on both **Huawei ECS** and **Azure VM**, utilizing scalable storage solutions like **Huawei OBS** and **Azure Data Lake**. --- ## 🎥 Project Demo | **Arabic Demonstration** | **English Demonstration** | |:---:|:---:| | 🎬 Watch Video | 🎬 Watch Video | --- ## 🧐 About the Project ### The Problem Roads are the most widely used transportation method globally. However, road distresses (cracks, potholes) significantly degrade ride quality and motorist safety [1]. * **Manual Inspection:** Traditional visual observation is inconsistent, time-consuming, and prone to human error [3]. * **High Cost:** Specialized quantitative analysis machines are prohibitively expensive. * **Maintenance Delays:** Lack of timely data leads to traffic congestion and costly emergency repairs [2]. ### The Solution: PavementEye We leverage **Data Science** and **Data Engineering** to transform road inspection into an automated intelligence pipeline. 1. **Ingestion:** Automatically collects streaming images from road cameras. 2. **Detection:** Uses **YOLOv8** deep learning models to detect and classify cracks. 3. **Enrichment:** Integrates with **OpenStreetMap** to tag distress points with precise geolocation data. 4. **Storage:** Archives raw images in a Data Lake (Cloud) and structured analytics in a NoSQL database. 5. **Visualization:** Presents a unified dashboard for administrators to view live road health and calculated **PCI (Pavement Condition Index)**. --- ## 🚀 Key Features * **⚡ Real-Time Streaming:** Processes data instantly using Apache Spark and Kafka for immediate anomaly detection. * **🧠 Advanced AI Analysis:** Fine-tuned Computer Vision models specifically for the *EGY_PDD* dataset. * **🌍 Geospatial Intelligence:** Maps specifi …