Open Source Geospatial Artificial Intelligece for societal applications

The increasing availability of open-source remote sensing datasets from satellite missions like Sentinel and Landsat has fundamentally transformed how we observe and respond to environmental and societal challenges. These datasets have opened new frontiers in transparency, reproducibility, and innovation. This talk introduces the foundational concepts behind open-source geospatial data and the FAIR principles (Findable, Accessible, Interoperable, Reusable) that govern their use, emphasizing how these principles support robust and collaborative geospatial analysis.

We will explore how satellite data has historically been used for environmental monitoring and how the rapid increase in data volume and temporal frequency has led to the rise of Geospatial Artificial Intelligence (GeoAI) (A combination of AI, remote sensing, and geographic information systems). GeoAI enables scalable, automated insights from petabytes of Earth observation data, moving from manual interpretation to real-time analysis. Participants will gain exposure to a suite of open-source tools such as PyTorch, TorchGeo, QGIS which will make researchers, developers, and practitioners to build, train, visualize, and deploy custom GeoAI models at scale.

One of the major bottlenecks in developing large-scale GeoAI models is the lack of high-quality labeled data. We will briefly discuss how recent advances in self-supervised learning are paving the way for label-agnostic geospatial foundation models. These models can be pre-trained on vast unlabeled datasets and fine-tuned for specific applications significantly reducing the dependence on expensive manual annotation.

Finally, the talk will highlight real-world case studies developed by Hydrosense Lab at IIT Delhi where open-source GeoAI has made tangible societal impact ranging from disaster response (e.g., floods, landslides), to crop monitoring and water resource management.

1. Understanding open-source remote sensing data like Sentinel, Landsat

2. FAIR principles (Findable, Accessible, Interoperable, Reusable)

3. How open source geospatial data has been used for understanding change and policy formulation

4. What is geospatial artificial intelligence GeoAI

5. How is GeoAI revolutionizing large scale geospatial data analysis

6. What are the open source tools for developing our own GeoAI models(eg pytorch, torchgeo)

7. Visualization of the products using Q-GIS.

8. Some of the examples of open source GeoAI for societal applications developed by hydrosense labs

9. Societal applications based using open source GeoAI outputs