FireRisk: A Remote Sensing Dataset for Fire Risk Assessment with Benchmarks Using Supervised and Self-supervised Learning
Shuchang Shen, Sachith Seneviratne, Xinye Wanyan, Michael Kirley
Abstract
In recent decades, wildfires have caused tremendous property losses, fatalities, and extensive damage to forest ecosystems. Inspired by the abundance of publicly available remote sensing projects and the burgeoning development of deep learning in computer vision, our research focuses on assessing fire risk using remote sensing imagery.In this work, we propose a novel remote sensing dataset, FireRisk, consisting of 7 fire risk classes with a total of 91872 labelled images for fire risk assessment. This dataset is labelled with the fire risk classes supplied by the Wildfire Hazard Potential (WHP) raster dataset [9], and remote sensing images are collected using the National Agriculture Imagery Program (NAIP) [27], a high-resolution remote sensing imagery program. On FireRisk, we present benchmark performance for supervised and self-supervised representations, with Masked Autoencoders (MAE) [16] pre-trained on ImageNet1k [8] achieving the highest classification accuracy, 65.29%.This remote sensing dataset, FireRisk, provides a new direction for fire risk assessment, and we make it publicly available on https://github.com/CharmonyShen/FireRisk.