Real-Time Aerial Fire Detection on Resource-Constrained Devices Using Knowledge Distillation

2025-02-28

AI Summary

To address the challenges of limited field-of-view and high computational overhead in real-time wildfire detection by edge devices (e.g., UAVs) deployed in large outdoor scenes, this paper proposes a lightweight aerial wildfire detection framework. Our method introduces knowledge distillation—applied jointly at both feature and logits levels—into the MobileViT-S architecture, significantly reducing model size and accelerating inference. We further incorporate Grad-CAM to enable interpretable, flame-region-focused visual verification. Additionally, we integrate edge-optimized inference techniques for deployment efficiency. Evaluated on mainstream wildfire benchmark datasets, our approach achieves accuracy improvements of 0.44% and 2.00% over state-of-the-art models, while drastically reducing parameter count and FLOPs. The optimized model attains >30 FPS inference speed on UAV hardware, demonstrating a favorable trade-off among high detection accuracy, strong interpretability, and practical edge deployability.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

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 learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Wildfire catastrophes cause significant environmental degradation, human losses, and financial damage. To mitigate these severe impacts, early fire detection and warning systems are crucial. Current systems rely primarily on fixed CCTV cameras with a limited field of view, restricting their effectiveness in large outdoor environments. The fusion of intelligent fire detection with remote sensing improves coverage and mobility, enabling monitoring in remote and challenging areas. Existing approaches predominantly utilize convolutional neural networks and vision transformer models. While these architectures provide high accuracy in fire detection, their computational complexity limits real-time performance on edge devices such as UAVs. In our work, we present a lightweight fire detection model based on MobileViT-S, compressed through the distillation of knowledge from a stronger teacher model. The ablation study highlights the impact of a teacher model and the chosen distillation technique on the model's performance improvement. We generate activation map visualizations using Grad-CAM to confirm the model's ability to focus on relevant fire regions. The high accuracy and efficiency of the proposed model make it well-suited for deployment on satellites, UAVs, and IoT devices for effective fire detection. Experiments on common fire benchmarks demonstrate that our model suppresses the state-of-the-art model by 0.44%, 2.00% while maintaining a compact model size. Our model delivers the highest processing speed among existing works, achieving real-time performance on resource-constrained devices.

Problem

• Real-time fire detection on resource-constrained devices
• Improving wildfire detection in remote and challenging areas
• Reducing computational complexity for edge devices like UAVs

Innovation

• Lightweight fire detection using MobileViT-S
• Knowledge distillation from stronger teacher model
• Real-time performance on resource-constrained devices