The escalating frequency and intensity of wildfires worldwide pose a growing challenge, highlighting the urgent need for comprehensive strategies to address and mitigate the impact of this environmental crisis. Wildfire management and prevention demands real-time spatial information for situational awareness. In response to the increasing need for timely spatial information to enhance wildfire management efforts, there has been a concerted effort to create practical Fire Management Decision Support (FMDS) systems throughout the world. A typical wildfire is a complex physical phenomenon and addressing them effectively requires a response that considers and incorporates different aspects or approaches. There is an unmet need for a complete set of AI-based image and video analysis tools for FMDS systems including UAV images and video, airplane imagery, and satellite images obtained in visible, infrared, and hyperspectral ranges. We will develop a set of real-time image and video-processing AI software for both wildfire prevention and management. Utilizing deep learning algorithms, our software will seamlessly integrate geospatial technologies, human activity data, and advanced computer vision tools. This integration aims to provide indispensable information to first responders, streamlining resource allocation, monitoring environmental conditions for enhanced fire prevention, and offering short-term fire forecasting capabilities to support more informed and effective management decisions. We pioneered image and video-based wildfire detection [1]- [5], however a complete image and video processing software package for FMDS systems does not exist.

We propose to develop the following vision-AI software suite covering pre-fire, active-fire, and post-fire problems:

• Powerline and nearby vegetation monitoring software using Low Earth Orbit (LEO) satellite,
  aircraft, and Unmanned Aerial Vehicle (UAV) based imagery (pre-fire).
• Real-time wildfire front and ember tracking software for UAVs and aircrafts (active-fire).
• Real-time first responder tracking and monitoring software for UAVs (active-fire).
• Video and image-based wildfire detection software for fixed cameras, drones, and other aerial
  sources including satellite and aircraft images (pre & active-fire).
• Generative AI for fire severity classification maps (post-fire).

Our software suite will integrate geospatial technologies, computer vision, and deep learning methods to deliver crucial information to first responders and decision makers. This information is used to facilitate effective resource allocation, monitoring of changing conditions to improve fire prevention, fire detection and monitoring, short-time fire forecasting (with ember detection), and post-fire analysis.

Our team consists of Prof. A. Enis Cetin, who was the technical lead in European Union funded FIRESENSE project during 2009-2013 [2], Ph.D. students Emadeldeen Hamdan and Shuaiang Rong, and four undergraduate electrical and computer engineering students. University of Illinois Chicago (UIC) is a Hispanic Serving Institute. We have the necessary lab facilities to develop the proposed software suite.