Measuring Vegetation Health to Predict Natural Hazards
summary
Uses Near Infrared (NIR) bands from Landsat satellite imagery on AWS to estimate vegetation health (chlorophyll content) and predict forest-fire likelihood; combines those predictions with MLS property data to assess risk to residential properties. Builds TensorFlow deep-learning models and deploys streaming preprocessing, training and inference pipelines using Apache Beam (Python) with the Flink runner and TFX for real-time, scalable predictions.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:58Introduction
- 4:25Landsat8 Payload
- 5:50How is this related to Vegetation? Photosynthetically Active Radiation (PAR) 400-700nm range
- 8:44Normalized Difference Vegetation Index (NDVI) Used to monitor agriculture production, drought, hazardous fire zones
- 15:42Filter Clouds
- 24:05How to Scale - Batch or Stream?
- 26:11Beam Inference Pipeline
related talks
48mTaking the planet’s pulse: AI to empower adaptation to global change
27mSmart Forestry: AI in the Service of Nature
36mWatching Millions of Trees with Python: For Impact and Profit
28mGeoAI: Evolution of AI-Based Tree/Vegetation Detection Models Using Imagery
58mImproving disaster response with machine learning
10mUsing Machine Learning to Reduce World Hunger
42mUsing Generative AI and Foundation Models to Predict Above Ground Biomass
30mUsing AI to Spot Deforestation-related Cows on Satellite Images