Near-real time fires detection using satellite imagery in Sudan conflict

2025-12-08

AI Summary

This study addresses the critical latency in wildfire monitoring within Sudan’s conflict-affected regions. We propose a near-real-time fire detection framework leveraging Planet Labs’ 4-band optical satellite imagery and a lightweight deep learning model. The method integrates multi-scale feature extraction with adaptive thresholding to jointly identify active fire pixels and delineate burned areas. Evaluated across five representative conflict-induced fire events, it achieves substantial improvements over conventional baselines—+12.3% in recall and sub-2-pixel localization accuracy. Key contributions include: (i) empirical validation that 4-band data suffices for high-accuracy fire monitoring, with marginal gains from 8-band or multi-temporal inputs, thereby significantly reducing data acquisition and computational overhead; (ii) an end-to-end latency of under 15 minutes, coupled with high scalability and robustness to cloud cover and sensor noise. The framework establishes a deployable technical paradigm for humanitarian remote sensing response in armed conflict settings.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

The challenges of ongoing war in Sudan highlight the need for rapid moni- toring and analysis of such conflicts. Advances in deep learning and readily available satellite remote sensing imagery allow for near real-time monitor- ing. This paper uses 4-band imagery from Planet Labs with a deep learning model to show that fire damage in armed conflicts can be monitored with minimal delay. We demonstrate the effectiveness of our approach using five case studies in Sudan. We show that, compared to a baseline, the automated method captures the active fires and charred areas more accurately. Our re- sults indicate that using 8-band imagery or time series of such imagery only result in marginal gains.

Problem

• Detects fire damage in conflicts using satellite imagery
• Applies deep learning for near-real-time monitoring in Sudan
• Compares effectiveness of different satellite imagery bands

Innovation

• Deep learning model analyzes 4-band satellite imagery
• Detects active fires and charred areas with minimal delay
• Automated method outperforms baseline for conflict monitoring