Two-Stage Framework for Efficient UAV-Based Wildfire Video Analysis with Adaptive Compression and Fire Source Detection

2025-08-22

AI Summary

To address the challenge of real-time wildfire video analysis and ignition detection under stringent onboard computational constraints of unmanned aerial vehicles (UAVs), this paper proposes a lightweight two-stage collaborative framework. In the first stage, a policy-network-driven frame compression mechanism, augmented by future-frame-informed salient point prediction, efficiently eliminates redundant video segments. In the second stage, upon flame response triggering, a lightweight, modified YOLOv8 model performs precise fire source localization. This work introduces, for the first time, the integration of sequence-level salient point modeling with an adaptive compression-detection co-design architecture. Evaluations on FLAME, HMDB51, and Fire & Smoke datasets demonstrate that the framework maintains high classification accuracy (>92%), improves detection mAP by 3.7%, and reduces inference latency by 38% compared to baseline methods—achieving a significant balance between real-time performance and detection accuracy.

Technology Category

• Computer Vision: Video Understanding & Activity Analysis
• Machine Learning: Learning on the Edge & Model Compression
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Unmanned Aerial Vehicles (UAVs) have become increasingly important in disaster emergency response by enabling real-time aerial video analysis. Due to the limited computational resources available on UAVs, large models cannot be run independently for real-time analysis. To overcome this challenge, we propose a lightweight and efficient two-stage framework for real-time wildfire monitoring and fire source detection on UAV platforms. Specifically, in Stage 1, we utilize a policy network to identify and discard redundant video clips using frame compression techniques, thereby reducing computational costs. In addition, we introduce a station point mechanism that leverages future frame information within the sequential policy network to improve prediction accuracy. In Stage 2, once the frame is classified as "fire", we employ the improved YOLOv8 model to localize the fire source. We evaluate the Stage 1 method using the FLAME and HMDB51 datasets, and the Stage 2 method using the Fire & Smoke dataset. Experimental results show that our method significantly reduces computational costs while maintaining classification accuracy in Stage 1, and achieves higher detection accuracy with similar inference time in Stage 2 compared to baseline methods.

Problem

• Efficient real-time wildfire video analysis on UAVs
• Adaptive compression to reduce computational costs
• Accurate fire source detection with lightweight models

Innovation

• Two-stage framework with adaptive video compression
• Policy network with future frame information
• Improved YOLOv8 model for fire localization