Forest fire surveillance systems: A review of deep learning methods

Saleh, Azlan; Zulkifley, Mohd Asyraf; Harun, Hazimah Haspi; Gaudreault, Francis; Davison, Ian; Spraggon, Martin

doi:10.1016/j.heliyon.2023.e23127

Cited by 26 publications

(4 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Some of the machine-learning and artificial intelligence techniques include support vector machine classifier, random forests, multilayer perception neural network, neuro-fuzzy models, etc. [104][105][106][107][108][109]. These models have their advantages and disadvantages.…”

Section: Lightning-induced Wildfire Modeling Methodsmentioning

confidence: 99%

“…For instance, the support vector machine can use a small dataset to predict wildfire incidences and produce high-accuracy results [95]. Some researchers in the fire science community have also contended that the accuracy of the random forest model is higher [104]. That is, the random forest model can simulate complex interactions between input models, and it is efficient for large datasets.…”

Section: Lightning-induced Wildfire Modeling Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Lightning-Induced Wildfires: An Overview

Song,

Xu,

et al. 2024

Fire

View full text Add to dashboard Cite

Wildfire causes environmental, economic, and human problems or losses. This study reviewed wildfires induced by lightning strikes. This review focuses on the investigations of lightning mechanisms in the laboratory. Also, the paper aims to discuss some of the modeling studies on lightning-induced wildfires at different geographical locations using satellite-recorded lightning data and different statistical analyses. This review established that irrespective of the different models used to predict lightning wildfires, there is still a lack of understanding of the lightning-strike ignition mechanism; few experiments have been modeled to establish the dynamics of lightning-strike ignition. Therefore, further research needs to be carried out in this area to understand lightning ignition. It was ascertained from the various statistical modeling that lightning-induced wildfires are exacerbated by the abundant availability of fuel with a lower moisture content and high lightning efficiency. Moreover, because of changes in the climate and weather conditions, i.e., harsh weather and climate conditions due to anthropogenic activities, lightning-induced ignition wildfires have increased over the years, and they are expected to increase in the future if the climate and weather conditions continue to aggravate. Although various modeling studies have identified that lightning-induced wildfires have increased recently, no preventive measures have been conclusively proposed to reduce lightning-caused wildfires. Hence, this aspect of research has to be given critical attention. This review presents information that gives a profound understanding of lightning-induced wildfires, especially factors that influence lightning wildfires, and the state-of-the-art research that has been completed to understand lightning-induced wildfires.

show abstract

Section: Lightning-induced Wildfire Modeling Methodsmentioning

confidence: 99%

Section: Lightning-induced Wildfire Modeling Methodsmentioning

confidence: 99%

Lightning-Induced Wildfires: An Overview

Song,

Xu,

et al. 2024

Fire

View full text Add to dashboard Cite

show abstract

“…Several studies have been presented recently to detect fire and smoke [23][24][25]. Talaat et al [26] proposed an improved fire detection approach for smart cities based on the YOLOv8 object detector algorithm; their approach is called the smart fire detection system (SFDS).…”

Section: Introductionmentioning

confidence: 99%

Fire and Smoke Detection Using Fine-Tuned YOLOv8 and YOLOv7 Deep Models

Chetoui,

Akhloufi

2024

Fire

View full text Add to dashboard Cite

Viewed as a significant natural disaster, wildfires present a serious threat to human communities, wildlife, and forest ecosystems. The frequency of wildfire occurrences has increased recently, with the impacts of global warming and human interaction with the environment playing pivotal roles. Addressing this challenge necessitates the ability of firefighters to promptly identify fires based on early signs of smoke, allowing them to intervene and prevent further spread. In this work, we adapted and optimized recent deep learning object detection, namely YOLOv8 and YOLOv7 models, for the detection of smoke and fire. Our approach involved utilizing a dataset comprising over 11,000 images for smoke and fires. The YOLOv8 models successfully identified fire and smoke, achieving a mAP:50 of 92.6%, a precision score of 83.7%, and a recall of 95.2%. The results were compared with a YOLOv6 with large model, Faster-RCNN, and DEtection TRansformer. The obtained scores confirm the potential of the proposed models for wide application and promotion in the fire safety industry.

show abstract

“…No obstante, la eficacia de estos modelos depende de la calidad y diversidad de los datos de entrenamiento (Diez, Kentsch, Fukuda, Caceres, Moritake & Cabezas, 2021). La variabilidad de los entornos donde ocurren los incendios plantea desafíos únicos para los modelos de DL, que deben ser entrenados con datos representativos para reconocer efectivamente entre situaciones benignas y peligrosas y al mismo tiempo evitar falsas alarmas o detecciones erróneas (Saleh, Zulkifley, Harun, Gaudreault, Davison & Spraggon, 2024). Para esto, se han propuesto enfoques multimodal que utilizan varios modos de datos como RGB, térmicos, hiperespectrales, tecnología láser infrarroja y datos meteorológicos para una detección robusta utilizando DL (Allison, Johnston, Craig & Jennings, 2016).…”

Section: Introductionunclassified

Research advances in Graphic and Design Engineering at the UPC BarcelonaTech

Farrerons Vidal,

Olmedo Torre,

Roncero Vivero

et al. 2024

View full text Add to dashboard Cite

Aquest llibre recull part de la recerca que porta a terme el Departament d’Enginyeria Gràfica i de Disseny (DEGD) de la Universitat Politècnica de Catalunya. Com podreu llegir, és una recerca molt variada i pluridisciplinària adaptada als temps actuals, on apareixen diverses de les principals preocupacions de la comunitat científica internacional, i referenciada en algunes de les línies d’investigació de Catalunya. En el llibre trobareu tan sols una petita mostra de la temàtica dels articles publicats pels professors del DEGD en les principals revistes científiques internacionals, però també en congressos, capítols de llibre, catàlegs d’exposició, projectes competitius industrials, projectes no competitius, projectes d’innovació docent i altres projectes de propietat intel·lectual industrial. El DEGD, a més de la recerca, té assignada docència a l’Escola Superior d’Enginyeries Industrial Aeroespacial i Audiovisual de Terrassa, a l’Escola Tècnica Superior d'Enginyeria Industrial de Barcelona, a l’Escola d'Enginyeria de Barcelona Est, a l’Escola Politècnica Superior d'Enginyeria de Vilanova i la Geltrú, a l’Escola Politècnica Superior d'Enginyeria de Manresa i a l’Escola d'Enginyeria de Telecomunicació i Aeroespacial de Castelldefels. Imparteix docència en gairebé totes les titulacions que assignades en aquestes escoles, tant en graus com en màsters universitaris. A més, els professors del departament col·laboren en diversos programes de doctorat. En tot cas, la docència és una branca més des de la que s’accedeix a la recerca.

show abstract

Forest fire surveillance systems: A review of deep learning methods

Cited by 26 publications

References 45 publications

Lightning-Induced Wildfires: An Overview

Lightning-Induced Wildfires: An Overview

Fire and Smoke Detection Using Fine-Tuned YOLOv8 and YOLOv7 Deep Models

Research advances in Graphic and Design Engineering at the UPC BarcelonaTech

Contact Info

Product

Resources

About