WildfireVLM: AI Framework for Early Wildfire Detection Using Satellite Imagery
Researchers have introduced WildfireVLM, an AI framework that combines satellite imagery with language-driven risk assessment for early wildfire detection. The system uses YOLOv12 to detect fire zones and smoke plumes, and integrates Multimodal Large Language Models (MLLMs) to convert detection outputs into contextualized risk assessments. The dataset is constructed using imagery from Landsat-8/9, GOES-16, and other Earth observation sources. The work addresses challenges such as faint smoke signals, dynamic weather, and the need for real-time analysis over large areas. The paper is available on arXiv under reference 2602.13305.
Key facts
- WildfireVLM is an AI framework for early wildfire detection and risk assessment.
- It uses satellite imagery from Landsat-8/9, GOES-16, and other sources.
- YOLOv12 is employed to detect fire zones and smoke plumes.
- Multimodal Large Language Models (MLLMs) provide contextualized risk assessment.
- The dataset includes labeled wildfire and smoke imagery.
- The system addresses challenges like faint smoke signals and dynamic weather.
- The paper is published on arXiv with ID 2602.13305.
- Wildfires are increasing due to climate change and human activities.
Entities
Institutions
- arXiv