Exploring State-of-the-art models for Early Detection of Forest Fires

TL;DR

This paper introduces a new synthetic dataset for early forest fire detection using visual analysis, and compares the performance of YOLOv7 and detection transformers on this dataset.

Contribution

It creates a novel dataset for early fire detection and evaluates state-of-the-art detection models on it, addressing data scarcity issues.

Findings

YOLOv7 outperforms detection transformers in early fire detection tasks.

Synthetic data from game simulators effectively complements real images.

Early detection of forest fires can be improved with tailored deep learning models.

Abstract

There have been many recent developments in the use of Deep Learning Neural Networks for fire detection. In this paper, we explore an early warning system for detection of forest fires. Due to the lack of sizeable datasets and models tuned for this task, existing methods suffer from missed detection. In this work, we first propose a dataset for early identification of forest fires through visual analysis. Unlike existing image corpuses that contain images of wide-spread fire, our dataset consists of multiple instances of smoke plumes and fire that indicates the initiation of fire. We obtained this dataset synthetically by utilising game simulators such as Red Dead Redemption 2. We also combined our dataset with already published images to obtain a more comprehensive set. Finally, we compared image classification and localisation methods on the proposed dataset. More specifically we used YOLOv7 (You Only Look Once) and different models of detection transformer.