Search for: All records

RGB imagery collected via Unmanned Aerial Systems (UAS) serves as a valuable resource for segmenting wildland fire scenes. However, collecting and labeling such data for deep learning–based image segmentation is both labor-intensive and computationally costly, and few reliable public datasets are available. To address this gap, we introduce BURN 1, a dataset developed for multiclass semantic segmentation of ash, fire, vegetation and background (non-burnable) pixels in wildland fire RGB imagery, intended for use with deep learning models such as the U-Net. 

Transformer-based models are popular for time series forecasting and spatiotemporal prediction due to their ability to infer semantic correlations in long sequences. However, for human mobility prediction, temporal correlations, such as location patterns at the same time on previous days or weeks, are essential. While positional encodings help retain order, the self-attention mechanism causes a loss of temporal detail. To validate this claim, we used a simple approach in the 2nd ACM SIGSPATIAL Human Mobility Prediction Challenge, predicting locations based on past patterns weighted by reliability scores for missing data. Our simple approach was among the top 10 competitors and significantly outperformed the Transformer-based model that won the 2023 challenge. 

The BURN 2 dataset is an extension of the previously published BURN 1 wildland fire dataset. BURN 1 includes synthetically generated images that include randomly pasted fire clusters from source to target images. In contrast, BURN 2 contains synthetically generated wildland fire images employing a contextually accurate data augmentation approach to generate realistic wildfire imagery. 

Human mobility data science using trajectories or check-ins of individuals has many applications. Recently, we have seen a plethora of research efforts that tackle these applications. However, research progress in this field is limited by a lack of large and representative datasets. The largest and most commonly used dataset of individual human trajectories captures fewer than 200 individuals, while datasets of individual human check-ins capture fewer than 100 check-ins per city per day. Thus, it is not clear if findings from the human mobility data science community would generalize to large populations. Since obtaining massive, representative, and individual-level human mobility data is hard to come by due to privacy considerations, the vision of this work is to embrace the use of data generated by large-scale socially realistic microsimulations. Informed by both real data and leveraging social and behavioral theories, massive spatially explicit microsimulations may allow us to simulate entire megacities at the person level. The simulated worlds, which do not capture any identifiable personal information, allow us to perform “in silico” experiments using the simulated world as a sandbox in which we have perfect information and perfect control without jeopardizing the privacy of any actual individual. In silico experiments have become commonplace in other scientific domains such as chemistry and biology, permitting experiments that foster the understanding of concepts without any harm to individuals. This work describes challenges and opportunities for leveraging massive and realistic simulated alternate worlds for in silico human mobility data science. 

Infectious disease spread within the human population can be conceptualized as a complex system composed of individuals who interact and transmit viruses through spatio-temporal processes that manifest across and between scales. The complexity of this system ultimately means that the spread of infectious diseases is difficult to understand, predict, and respond to effectively. Research interest in GeoAI for public health has been fueled by the increased availability of rich data sources such as human mobility data, OpenStreetMap data, contact tracing data, symptomatic online surveys, retail and commerce data, genomics data, and more. This data availability has resulted in a wide variety of data-driven solutions for infectious disease spread prediction which show potential in enhancing our forecasting capabilities. This book chapter (1) motivates the need for AI-based solutions in public health by showing the heterogeneity of human behavior related to health, (2) provides a brief survey of current state-of-the-art solutions using AI for infectious disease spread prediction, (3) describes a use-case of using large-scale human mobility data to inform AI models for the prediction of infectious disease spread in a city, and (4) provides future research directions and ideas.