AI-powered Flood Monitoring and Smart Navigation System for Ho Chi Minh City
🎓 Graduation Thesis from University of Information Technology - Vietnam National University Ho Chi Minh City
🏆 Final Score: 9.9 / 10
📰 "UIT-VisDrone-Flood: A Synthesized Aerial Vehicle Detection Dataset Under Flood Conditions"
📍 Published at: 13th International Conference on Control, Automation and Information Sciences (ICCAIS 2024)
📄 IEEE Xplore | PDF | AI Guide
Safe-Move is a real-time AI-integrated system designed to monitor urban flooding and support traffic navigation. The system empowers both citizens and local authorities with accurate, live information about flood conditions across the city.
- 🔍 600+ traffic cameras monitored in real time.
- 🧪 Created a synthetic dataset of 7,411 images simulating flood conditions.
- 🧠 AI flood detection accuracy: ~90%.
- 📲 Notification delay: < 5 seconds from detection to alert.
- 📡 12-second refresh on camera snapshots.
- CNN-based flood detection model using transfer learning.
- Real-time inference on camera feeds.
- Automatic flood zone classification.
- Alert users via push/email and sync to flood maps.
- AI Monitoring Service runs 24/7 as a long-living containerized service.
- Continuously analyzes traffic camera images.
- Detects flooded roads using CNN.
- Updates interactive flood maps.
- Triggers alerts to affected users.
👉 See AI Guide
- Built with microservices architecture, each responsible for a specific domain (flood detection, camera control, email, notification, authentication).
- Deployed and managed via Docker and
docker-compose. - Hosts core business logic, authentication, API routes.
- Interfaces with PostgreSQL, Redis, and external services.
👉 See Backend Installation Guide
- Built with ReactJS + TailwindCSS.
- Used by local authorities to manage camera devices.
- Review and verify citizen flood reports.
- Send manual alerts if necessary.
👉 See Web Admin Setup Guide
- Targeted at general users.
- Integrated with Google Maps SDK and HERE Maps API.
- Smart route planning feature to avoid flooded areas.
- Report flooding with images.
- View real-time flood maps and camera feeds.
- Receive notifications and reroute suggestions.
👉 See Mobile App Setup Guide
- ✅ Real-time monitoring of 600+ asynchronous camera feeds.
- ✅ Low-latency flood detection pipeline using AI.
- ✅ Image upload + classification + alert dispatch under 5s.
- ✅ Role-based access and permission for admin/user.
- ✅ Multi-platform deployment (Web, Mobile, API backend).
- ✅ Smart routing integrated with external map APIs.
- Supabase: Media storage (flood photos, camera snapshots).
- Firebase: User authentication.
- SendGrid: Email alerts.
- Render: App deployment.
- Neon DB: Managed PostgreSQL instance.
- 🎓 Paper: IEEE Xplore | PDF
- 🗃️ Dataset: UIT-Flooded-VisDrone on Roboflow
- 🧪 YOLOv10 Model Demo: Hugging Face Space
- Sep 2024: Planning, research, architecture.
- Oct–Nov 2024: Development of backend, mobile, AI.
- Dec 2024: Testing, deployment, and defense.
Developed by Hồ Đình Mạnh & Lê Thị Bích Hằng
Supervised by Dr. Nguyễn Tấn Trần Minh Khang & Dr. Nguyễn Duy Khánh