Hi everyone! ![]()
I’m excited to share my latest project built with Streamlit for the NASA Space Apps Challenge: SAR-MotionLens!
It’s an advanced web application designed for global Earth dynamics intelligence, allowing users to explore SAR interferometry (InSAR), land subsidence, fault creep, and geological anomalies interactively.
Key Features & Tech Stack:
- Multi-Sensor Data Integration: Supports NISAR (L/S-Band) & Sentinel-1 (C-Band) data along with PolSAR decomposition.
- Machine Learning Engine: Powered by LightGBM & Isolation Forest for high-hazard anomaly detection and regime classification (achieving ~99.38% 5-fold CV accuracy).
- Interactive Geospatial Visualization: Combines PyDeck and Plotly to let users click anywhere on the globe and analyze time-series kinematics.
- Guided Scenarios: Pre-configured presets like the Tehran-Shahriar plain subsidence to instantly inspect real-world deformation regimes.
Live App: https://sar-motionlens.streamlit.app/
I’d love for you to check it out, test it, and share your feedback or suggestions! Let me know if you have any questions about how it’s implemented using Streamlit or the underlying ML pipeline. ![]()