Apply responsible AI principles when building remote sensing datasets
summary
Dynamic World is a 10‑meter, near‑real‑time global land‑use/land‑cover dataset built with Sentinel‑2 imagery, Google Earth Engine and Vertex AI, trained on 24,000+ annotated scenes (CC BY‑4.0) and deployed with a cloud‑shadow mask and per‑pixel class probabilities. An ethics review defines allowed uses, gated Earth Engine access enforces a permission portal, and safeguards include clear licensing, terms of use, a reporting channel and tracking of users. Expect probability bands per pixel, real‑time derivative maps, and documented limitations and thresholds for responsible deployment.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:55Dynamic World
- 1:41Goal of Dynamic World Data Products
- 6:23Earth Engine Code Editor
- 9:41The Limitations of the Model
- 14:35Examining Model Limitations
- 18:10The User
- 18:58The Impacts
- 23:47Track Usage or Users
related talks
48mTaking the planet’s pulse: AI to empower adaptation to global change
27mSmart Forestry: AI in the Service of Nature
11mTerraKit: Standardising AI-Ready Geospatial Data...
31mThe Next Digital Twin is...Earth!
18mGeospatial AI: Where Can AI Help to Tackle Today’s Real-World Challenges?
09mGen AI Workflows for Climate Risk Modeling in Food Systems
17mFree satellite imagery and DEM analyses enabling NRM in the developing world
28mGeoAI: Evolution of AI-Based Tree/Vegetation Detection Models Using Imagery