Fire-Net: A Deep Learning Framework for Active Forest Fire Detection

Seyd Teymoor Seydi,Vahideh Saeidi,Bahareh Kalantar,Naonori Ueda,Alfian Abdul Halin

DOI: https://doi.org/10.1155/2022/8044390

IF: 2.336

2022-02-21

Journal of Sensors

Abstract:Forest conservation is crucial for the maintenance of a healthy and thriving ecosystem. The field of remote sensing (RS) has been integral with the wide adoption of computer vision and sensor technologies for forest land observation. One critical area of interest is the detection of active forest fires. A forest fire, which occurs naturally or manually induced, can quickly sweep through vast amounts of land, leaving behind unfathomable damage and loss of lives. Automatic detection of active forest fires (and burning biomass) is hence an important area to pursue to avoid unwanted catastrophes. Early fire detection can also be useful for decision makers to plan mitigation strategies as well as extinguishing efforts. In this paper, we present a deep learning framework called Fire-Net, that is trained on Landsat-8 imagery for the detection of active fires and burning biomass. Specifically, we fuse the optical (Red, Green, and Blue) and thermal modalities from the images for a more effective representation. In addition, our network leverages the residual convolution and separable convolution blocks, enabling deeper features from coarse datasets to be extracted. Experimental results show an overall accuracy of 97.35%, while also being able to robustly detect small active fires. The imagery for this study is taken from Australian and North American forests regions, the Amazon rainforest, Central Africa and Chernobyl (Ukraine), where forest fires are actively reported.

engineering, electrical & electronic,instruments & instrumentation

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the automatic detection of forest fires. Specifically, the author proposes a deep - learning - based method - Fire - Net, which is used to detect active forest fires and burning biomass from Landsat - 8 images. Whether forest fires occur naturally or are caused by humans, they can spread rapidly, causing great damage and loss of life. Therefore, timely and accurate detection of active forest fires is crucial for reducing or preventing these disasters. Early fire detection not only helps decision - makers formulate mitigation strategies, but also enables effective firefighting work.

The paper mentions that existing forest fire detection methods are mainly based on traditional machine - learning classification, which requires feature engineering and has challenges when facing changes in environmental conditions, air temperature fluctuations, and small - fire detection. To overcome these problems, the author designed Fire - Net, a two - stream - channel deep - learning framework that combines optical (red, green, blue) and thermal - modality images, using a convolutional neural network architecture to extract deep features, which is especially suitable for coarse - grained datasets. In addition, Fire - Net also adopts residual convolution and depth - wise separable convolution blocks to enhance the detection ability of small fires. Experimental results show that the overall accuracy of Fire - Net reaches 97.35% and it can robustly detect small active fires.

Overall, this research aims to improve the efficiency and accuracy of forest fire detection by proposing a new deep - learning framework, especially when using freely accessible data such as Landsat - 8 satellite images. This method can not only improve the detection ability of small fires, but also automatically and accurately detect active fires, and has the potential for real - time applications.