The Landslide Detection System is a cloud-based platform designed to identify and monitor potential landslide-prone areas using advanced machine learning models. The system integrates a backend powered by AI, a frontend interface for user interaction, and cloud infrastructure for seamless deployment, monitoring, and feedback. This project is developed as part of the CS462: Cloud Computing course to demonstrate end-to-end application development using DevOps principles.
- Machine Learning-Based Detection: Utilizes trained models to predict landslide risks based on environmental data.
- Continuous Feedback: Collects user and system feedback for iterative improvements.
- Cloud Deployment: Fully hosted on Google Cloud Run for scalability and reliability.
- Frontend: React,
- Backend: Python, Flask, TensorFlow
- Machine Learning: Trained models in .h5 format
- DevOps Tools: Docker, GitHub Actions, PyTest, Prometheus, Grafana
- Cloud Platform: Google Cloud Run
- Monitoring & Feedback: LogRocket, Grafana
Data Collection: Environmental data (e.g., rainfall, soil moisture, terrain features) is fed into the system. Model Prediction: The AI model processes the data and predicts the likelihood of landslides. Result Visualization: Predictions are displayed on an interactive web dashboard. System Monitoring: Continuous monitoring ensures system reliability, tracks user actions, and logs errors.
We created a .yml file that used Github Actions for integration, Docker for containerization and pytest for testing. The workflow is triggered when a file is pushed to "main"
