Research

My research is on the use of Deep Learning methods for spatiotemporal earth observation 🛰️ datasets, with a particular focus on the modeling of wildfires 🔥. In an attempt to promote openness and reproducibility, in this page you will find links to my papers, presentations and related code.

📜 Papers

For a full updated list, please refer to my scholar profile. For each paper, first authors are presented in bold.

• TeleViT: Teleconnection-driven Transformers Improve Subseasonal to Seasonal Wildfire Forecasting. Prapas, I., Bountos, N. I., Kondylatos, S., Michail, D., Camps-Valls, G., & Papoutsis, I. (2023). In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 3754-3759). (ICCV 2023, AI+HADR workshop, Best Paper Award 🥇). paper | code

• Mesogeos: A multi-purpose dataset for data-driven wildfire modeling in the Mediterranean. Kondylatos, S., Prapas, I., Camps-Valls, G., & Papoutsis, I. (2023). arXiv preprint arXiv:2306.05144 (Accepted at NeurIPS 2023 Datasets and Benchmarks Track). html | code

• Wildfire danger prediction and understanding with Deep Learning. Kondylatos, S., Prapas, I., Ronco, M., Papoutsis, I., Camps-Valls, G., Piles, M., … & Carvalhais, N. (2022). Geophysical Research Letters, 49(17), e2022GL099368. html | code

• Deep Learning for Rapid Landslide Detection using Synthetic Aperture Radar (SAR) Datacubes. Boehm, V., Leong, W. J., Mahesh, R. B., Prapas, I., Nemni, E., Kalaitzis, F., … & Ramos-Pollan, R. (2022). arXiv preprint arXiv:2211.02869. Presented at Tackling Climate Change with AI workshop in NeurIPS 2022. html | code

• Deep learning for global wildfire forecasting. Prapas, I., Ahuja, A., Kondylatos, S., Karasante, I., Panagiotou, E., Alonso, L., … & Papoutsis, I. (2022). arXiv preprint arXiv:2211.00534. Presented at Tackling Climate Change with AI workshop in NeurIPS 2022. html

• Deep learning methods for daily wildfire danger forecasting. Prapas, I., Kondylatos, S., Papoutsis, I., Camps-Valls, G., Ronco, M., Fernández-Torres, M. Á., … & Carvalhais, N. (2021). arXiv preprint arXiv:2111.02736. Presented at NeurIPS 2021 AI+HADR workshop. html | code

• Continuous training and deployment of deep learning models. Prapas, I., Derakhshan, B., Mahdiraji, A. R., & Markl, V. (2021). Datenbank-Spektrum, 21(3), 203-212. Presented at LWDA 2021. html | code

• Towards human activity reasoning with computational logic and deep learning. Prapas, I., Paliouras, G., Artikis, A., & Baskiotis, N. (2018, July). In Proceedings of the 10th Hellenic Conference on Artificial Intelligence (pp. 1-4). preprint | paper

💾 Datasets

• FireCube: A Daily Datacube for the Modeling and Analysis of Wildfires in Greece.

• SeasFire Cube: A Global Dataset for Seasonal Fire Modeling in the Earth System.

• Mesogeos: A multi-purpose dataset for data-driven wildfire modeling in the Mediterranean.

🎨 Presentations (invited talks, conferences, posters)

• TeleViT: Teleconnection-driven Transformers Improve Subseasonal to Seasonal Wildfire Forecasting. AI+HADR, ICCV 2023, Paris, France (Spotlight Talk, Best Paper Award 🥇). paper | poster

• Wildfire danger forecasting at different spatio-temporal scales. AI for Good 2023, ELLIS ML Workshop “The role of AI in tackling climate change and its impacts: from science to early warning”, Geneva, Switzerland (Invited Talk). presentation

• Earth System Deep Learning towards a Global Digital Twin of Wildfires. EGU 2023, Vienna, Austria (Oral Presentation). presentation | abstract

• Deep learning for wildfire danger forecasting at different spatio-temporal scales. ITU Webinar “Fighting wildfires with AI-powered insights” April 19th, 2023 (Invited Talk). presentation | recording

• AI4EO ESA Φ-Lab Workshop on AI for Natural Hazard Management, ESRIN Frascatti Italy, May 10th 2023 (Co-organized).

  • Deep Learning for Landslide Detection (Oral Presentation). presentation

  • Deep Learning for Wildfire Danger Forecasting at Different Spatiotemporal Scales (Oral Presentation). presentation

• Earth System Deep Learning for Global Wildfire Forecasting. ML4ESOP, ECMWF Reading, United Kingdom (Poster presentation). poster

• Self Supervised Learning on SAR Data For Change Detection. NASA Challenge, FDL USA, September 15th 2022 (Online Presentation). presentation

• Deep Learning Methods for Daily Wildfire Danger Forecasting. Living Planet Symposium 2022, Bonn, Germany (Poster Presentation). poster

• Deep Learning Methods for Daily Wildfire Danger Forecasting. Poster in hadr.ai Workshop, NeurIPS 2021 (Online Presentation). poster | paper

• Continuous Training and Deployment of Deep Learning Models. Presented in LWDA workshop 2021 (Online Presentation). preprint | paper | presentation

🏛️ Tutorials

• Deep Learning for monitoring and forecasting natural hazards with earth observation data. IGARSS Tutorial (2023). github repo

If you are searching for a resource with a broken link, please drop me an email