Preprint / Version 1

Satellite Imagery Data Curation Workflow for Wildfire Detection With Advanced Segmentation Modeling


  • Anjali Singh Saratoga High School



Artificial intelligence, Machine learning, Satellite imagery, Wildfire, Segmentation, U-Net, ViT


Across the globe, wildfires are occurring at increased frequency, significantly impacting ecosystems and human civilizations. This research paper focuses on the efficacy of two of the most advanced semantic segmentation machine learning models, specifically U-Net based on convolutional neural network and SegFormer based on Vision Transformer network for wildfire detection utilizing Moderate Resolution Imaging Spectroradiometer (MODIS) satellite imagery. The dataset is assembled using a specially built pipeline composed of 1) a workflow to obtain the wildfire candidate location and date, and 2) a subsequent step to collect satellite imagery data utilizing Google Earth Engine image collection and image download service. Experimental evaluation on this dataset shows that both models demonstrate high predictive power of fire at specific geolocations, with ViT outperforming U-Net at the edges of fire regions.


Statistics. (2022, October 24). California Department of Forestry and Fire Protection. Retrieved January 1, 2024, from

National Interagency Coordination Center Wildland Fire Summary and Statistics Annual Report 2022. (2022). National Interagency Fire Center. Retrieved January 1, 2024, from %20Reports/2022/annual_report.2.pdf

Binskin, M. (2020, October 28). Royal Commission into National Natural Disaster Arrangements Report. Royal Commission into Natural Disaster Arrangements.

Hughes, L., Dean, A., Steffen, W., Rice, M., & Mullins, G. (2020, November 3). Summer of crisis. Climate Council. Retrieved December 25, 2023, from

Pham, K., Ward, D., & Rubio, S. (2023). California Wildfire Prediction using Machine Learning. 2022 21st IEEE International Conference on Machine Learning and Applications (ICMLA).

Liu, Y., Le, S., & Zou, Y. (2023). A simplified machine learning based wildfire ignition model from insurance perspective. ICLR 2023 Workshop on Tackling Climate Change with Machine Learning.

Haim, Z. B., & Navo, O. (2023, February 3). Real-time tracking of wildfire boundaries using satellite imagery. Google Research.

Thisanke, H., Deshan, C., & Chamith, K. (2023, May 5). Semantic Segmentation using Vision Transformers: A survey.

Ronneberger, O., Fischer, P., & Brox, T. (2015, May 18). U-Net: Convolutional Networks for Biomedical Image Segmentation.

Xie, E., Wang, W., & Yu, Z. (2021, May 31). SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers.

Giglio, L., Justice, C. (2021). MODIS/Terra Thermal Anomalies/Fire Daily L3 Global 1km SIN Grid V061 [Data set]. NASA EOSDIS Land Processes Distributed Active Archive Center. Retrieved January 1, 2024, from

Giglio, L., Schroeder, W., & Hall, J. V. (2021, May). MODIS Collection 6 and Collection 6.1 Active Fire Product User's Guide Version 1.0. MODIS Active Fire Products. Retrieved January 1, 2024, from

Google Earth Engine (2023). A planetary-scale platform for Earth science data & analysis.

Chollet, F. (2019, March 20). Image segmentation with a U-Net-like architecture. Keras.

Paul, S. (2023, January 25). Semantic segmentation with SegFormer and Hugging Face Transformers. Keras. Retrieved December 25, 2023, from