Fire Dataset Research Articles

Modeling actuarial data using iterated trigonometric distributions

. A new class of trigonometric distribution (TD) is developed iteratively by replacing the random variable of the TD function (or its survival function) with another TD function (or its survival function). The resulting families of distributions produce various density and hazard shapes and a wide variety of tail shapes for the parametric modeling of complex data arising in actuarial science. The basic properties of these distributions are studied and presented. The new TDs successfully model the AON Re Belgian and Danish fire insurance data sets. Risk measures such as value at risk (VaR) and conditional tail expectation (CTE) are computed and compared with their empirical values and other distributions presented in actuarial literature. Also, limited expected values (LEV) for various policy limits are computed to compare with empirical counterparts to emphasize the new TDs’ validity for actuarial predictions.

FF-net: A Target Detection Method Tailored for Mid-to-Late Stages of Forest Fires in Complex Environments

To address the suboptimal detection of forest fires in complex environments during the middle and late stages, we have developed a novel object detection algorithm, FF-net (F_Res, Fire Label Assignment). Initially, a novel fire extraction method, F_Res (Fire ResNet), was designed, incorporating an F_fire activation function to enhance the network's feature extraction and nonlinear fitting capabilities. To reduce the complexity of the network while ensuring detection accuracy, a Fire Label Assignment (FLA) method was devised. Ultimately, the Kullback-Leibler (KL) divergence was integrated with the Focal Loss to create a loss function, KLF (Kullback-Leibler Focal Loss), capable of addressing data imbalance and preventing gradient vanishing and explosion. We reconstructed aerial forest fire and flame datasets and, employing 16 evaluative metrics, compared experiments with classical CNN algorithms, SOTA (State-of-the-Art) algorithms, Transformer algorithms, and advanced fire detection algorithms, discovering that the overall detection efficacy of FF-net surpasses that of other algorithms. In terms of precision on the fire dataset, it exceeds the performance of classical CNN algorithms, SOTA algorithms, and Transformer algorithms by 6.46%, 6.40%, and 21.50% respectively. On the flame dataset, the mean accuracy is 2.47% higher than that of the second-ranking algorithm. Subsequently, visualizing the network's performance on both datasets using Grad-CAM once again confirmed the suitability of FF-net for the task of aerial forest fire detection. The source code of the paper is publicly available at: https://github.com/jyyuan666/FF-net.

Increasing corn cultivation exacerbated crop residue burning in Northeast China in 21st century

China's endeavors to mitigate recurrent crop residue burning (CRB) and improve air quality have yielded positive results owing to recent pollution prevention policies. Nonetheless, persistent challenges remain, particularly in the Northeast China (NEC), where low temperature complicates crop residue management. Here, we examined the effects of cropping pattern adjustment on variations of CRB patterns in NEC during 2001–2021, utilizing the Moderate-resolution Imaging Spectroradiometer (MODIS) burned area dataset, the Visible Infrared Imaging Radiometer Suite (VIIRS) active fire dataset, and the high-accuracy crop planting area maps. Our results revealed an overall upward trend of 805.96 km2/yr in NEC CRB from 2001 to 2021. The corn CRB area accounted for more than 50% of the total CRB area in each CRB-intensive year (2013–2021), and the increasing corn CRB generally aligns with the growing corn cultivation fields. A seasonal shift in CRB was found around 2017, with intensive CRB activities transitioning from both autumn and spring to primarily spring, particularly in the Songnen Plain and Sanjiang Plain. The changing trend of PM2.5 concentration aligned spatially with the shift. Moreover, the CRBs in spring of 2020 and 2021 were more severe than the major burning seasons in previous years, likely due to the disruptions during COVID-19 lockdowns. In certain years, the explanatory power of spring CRB on PM2.5 concentration was comparable to that of other natural factors, such as precipitation. This study underscores the critical need for sustained and region-specific strategies to tackle the challenges posed by CRBs.

FireDA: A Domain Adaptation-Based Method for Forest Fire Recognition with Limited Labeled Scenarios

Vision-based forest fire detection systems have significantly advanced through Deep Learning (DL) applications. However, DL-based models typically require large-scale labeled datasets for effective training, where the quality of data annotation is crucial to their performance. To address challenges related to the quality and quantity of labeling, a domain adaptation-based approach called FireDA is proposed for forest fire recognition in scenarios with limited labels. Domain adaptation, a subfield of transfer learning, facilitates the transfer of knowledge from a labeled source domain to an unlabeled target domain. The construction of the source domain FBD is initiated, which includes three common fire scenarios: forest (F), brightness (B), and darkness (D), utilizing publicly available labeled data. Subsequently, a novel algorithm called Neighborhood Aggregation-based 2-Stage Domain Adaptation (NA2SDA) is proposed. This method integrates feature distribution alignment with target domain Proxy Classification Loss (PCL), leveraging a neighborhood aggregation mechanism and a memory bank designed for the unlabeled samples in the target domain. This mechanism calibrates the source classifier and generates more accurate pseudo-labels for the unlabeled sample. Consequently, based on these pseudo-labels, the Local Maximum Mean Discrepancy (LMMD) and the Proxy Classification Loss (PCL) are computed. To validate the efficacy of the proposed method, the publicly available forest fire dataset, FLAME, is employed as the target domain for constructing a transfer learning task. The results demonstrate that our method achieves performance comparable to the supervised Convolutional Neural Network (CNN)-based state-of-the-art (SOTA) method, without requiring access to labels from the FLAME training set. Therefore, our study presents a viable solution for forest fire recognition in scenarios with limited labeling and establishes a high-accuracy benchmark for future research.

Dehazing Algorithm Integration with YOLO-v10 for Ship Fire Detection

Ship fire detection presents significant challenges in computer vision-based approaches due to factors such as the considerable distances from which ships must be detected and the unique conditions of the maritime environment. The presence of water vapor and high humidity further complicates the detection and classification tasks for deep learning models, as these factors can obscure visual clarity and introduce noise into the data. In this research, we explain the development of a custom ship fire dataset, a YOLO (You Only Look Once)-v10 model with a fine-tuning combination of dehazing algorithms. Our approach integrates the power of deep learning with sophisticated image processing to deliver comprehensive solutions for ship fire detection. The results demonstrate the efficacy of using YOLO-v10 in conjunction with a dehazing algorithm, highlighting significant improvements in detection accuracy and reliability. Experimental results show that the YOLO-v10-based developed ship fire detection model outperforms several YOLO and other detection models in precision (97.7%), recall (98%), and mAP@0.50 score (89.7%) achievements. However, the model reached a relatively lower score in terms of F1 score in comparison with YOLO-v8 and ship-fire-net model performances. In addition, the dehazing approach significantly improves the model’s detection performance in a haze environment.

LD-YOLO: A Lightweight Dynamic Forest Fire and Smoke Detection Model with Dysample and Spatial Context Awareness Module

The threat of forest fires to human life and property causes significant damage to human society. Early signs, such as small fires and smoke, are often difficult to detect. As a consequence, early detection of smoke and fires is crucial. Traditional forest fire detection models have shortcomings, including low detection accuracy and efficiency. The YOLOv8 model exhibits robust capabilities in detecting forest fires and smoke. However, it struggles to balance accuracy, model complexity, and detection speed. This paper proposes LD-YOLO, a lightweight dynamic model based on the YOLOv8, to detect forest fires and smoke. Firstly, GhostConv is introduced to generate more smoke feature maps in forest fires through low-cost linear transformations, while maintaining high accuracy and reducing model parameters. Secondly, we propose C2f-Ghost-DynamicConv as an effective tool for increasing feature extraction and representing smoke from forest fires. This method aims to optimize the use of computing resources. Thirdly, we introduce DySample to address the loss of fine-grained detail in initial forest fire images. A point-based sampling method is utilized to enhance the resolution of small-target fire images without imposing an additional computational burden. Fourthly, the Spatial Context Awareness Module (SCAM) is introduced to address insufficient feature representation and background interference. Also, a lightweight self-attention detection head (SADH) is designed to capture global forest fire and smoke features. Lastly, Shape-IoU, which emphasizes the importance of boundaries’ shape and scale, is used to improve smoke detection in forest fires. The experimental results show that LD-YOLO realizes an mAP0.5 of 86.3% on a custom forest fire dataset, which is 4.2% better than the original model, with 36.79% fewer parameters, 48.24% lower FLOPs, and 15.99% higher FPS. Therefore, LD-YOLO indicates forest fires and smoke with high accuracy, fast detection speed, and a low model complexity. This is crucial to the timely detection of forest fires.

Measuring dimensions of teacher resilience in Africa: Self-efficacy and teacher efficacy

With this article we aimed to contribute evidence on reliable and valid measurement of teacher resilience in an under researched African context and population. Scales from an existing instrument, ENTREEi, were used to measure the resilience of pre-service teachers at a South African university. The sample constituted 1,193 final-year pre-service teachers (20–32 years), who completed the FIRE teacher resilience measure (2015 to 2017). Teacher resilience data were purposively selected from the FIRE dataset and items analysed related to self-efficacy and teacher efficacy beliefs. A quantitative approach was used, which consisted of reliability (internal consistency using Cronbach’s alpha) and validity (construct validity using Spearman correlations) analyses of the teacher resilience scale and the teacher efficacy scale. The statistical analyses indicate that the FIRE teacher resilience measure is a reliable and valid measure for intrapersonal resilience-enabling pathways in a challenged context. The underlying variable structure of the teacher resilience scale and teacher efficacy scale held in a Global South setting when used with pre-service teachers. The article contributes to teacher resilience measurement discourse by providing insights into the utility of teacher resilience scales in South Africa. The results act as a precursor for comparative teacher resilience results worldwide for future studies.

Predicting Forest Fire Area Growth Rate Using an Ensemble Algorithm

Due to its unique geographical and climatic conditions, the Liangshan Prefecture region is highly prone to large fires. There is an urgent need to study the growth rate of fire-burned areas to fill the research gap in this region. To address this issue, this study uses the Grey Wolf Optimizer (GWO) algorithm to optimize the hyperparameters in the eXtreme Gradient Boosting (XGBoost) model, constructing a GWO-XGBoost model. Finally, the optimized ensemble model (GWO-XGBoost) is used to create a fire growth rate warning map for the Liangshan Prefecture in Sichuan Province, China, filling the research gap in forest fire studies in this area. This study comprehensively selects factors such as monthly climate, monthly vegetation, terrain, and socio–economic aspects and incorporates monthly reanalysis data from forest fire assessment systems in Canada, the United States, and Australia as features to construct the forest fire dataset. After collinearity tests to filter redundant features and Pearson correlation analysis to explore features related to the burned area growth rate, the Synthetic Minority Oversampling Technique (SMOTE) is used to oversample the positive class samples. The GWO algorithm is used to optimize the hyperparameters in the XGBoost model, constructing the GWO-XGBoost model, which is then compared with XGBoost, Random Forest (RF), and Logistic Regression (LR) models. Model evaluation results showed that the GWO-XGBoost model, with an AUC value of 0.8927, is the best-performing model. Using the SHapley Additive exPlanations (SHAP) value analysis method to quantify the contribution of each influencing factor indicates that the Ignition Component (IC) value from the United States National Fire Danger Rating System contributes the most, followed by the average monthly temperature and the population density. The growth rate warning map results indicate that the southern part of the study area is the key fire prevention area.

Assessment of retinal blood vessel segmentation using U-Net model: A deep learning approach

Segmentation of retinal blood vessels from fundus images is vital to assist ophthalmologists in diagnosing different eye diseases like Arteriosclerosis, Glaucoma, Diabetic Retinopathy, Hypertension, and Choroidal Neovascularization. Accurate detection and timely treatment of these eye diseases can prevent irreversible impairment of vision. In computer-aided automatic diagnosis, Preprocessing of fundus images is necessary to reduce the uneven retinal image illumination, as well as the undesirable effect arising from the contrast differences between the retinal blood vessels and the background. In this study, the Hessian-based Frangi vesselness filter is proposed to enhance vasculature in fundus images. This approach seeks to extract thin vessels to diminish the intensity disparity between thick and thin vessels. Its accuracy has never, however, been thoroughly evaluated. To verify the effectiveness of the suggested filter for boosting vessel-like structures in the fundus images, an experimental approach is presented in this paper. The ability of the filter to improve vessel-like structures in fundus images is validated, and an experimental technique for evaluating filter performance using the U-Net model is described. Five quantitative performances, namely Accuracy, Recall, Specificity, Precision, and F1 Score measures on the DRIVE, HRF, DIARETDB1, DIARETDB0, CHASEDB1, ORIGA, DRISHTI GS, DRIONS DB, FIRE, and FIOT datasets, were evaluated and quantified in these experiments to validate the efficacy of the proposed filter. The results demonstrated a noteworthy improvement by achieving an Accuracy of 99.79 %, 99.84 %, 99.64 %, 99.72 %, 99.57 %, 99.48 %, 99.77 %, 99.05 %, 99.93 %, and 99.40 % respectively. Empirical evidence supports the advantages of this approach.

Real-Time Wildfire Monitoring Using Low-Altitude Remote Sensing Imagery

With rising global temperatures, wildfires frequently occur worldwide during the summer season. The timely detection of these fires, based on unmanned aerial vehicle (UAV) images, can significantly reduce the damage they cause. Existing Convolutional Neural Network (CNN)-based fire detection methods usually use multiple convolutional layers to enhance the receptive fields, but this compromises real-time performance. This paper proposes a novel real-time semantic segmentation network called FireFormer, combining the strengths of CNNs and Transformers to detect fires. An agile ResNet18 as the encoding component tailored to fulfill the efficient fire segmentation is adopted here, and a Forest Fire Transformer Block (FFTB) rooted in the Transformer architecture is proposed as the decoding mechanism. Additionally, to accurately detect and segment small fire spots, we have developed a novel Feature Refinement Network (FRN) to enhance fire segmentation accuracy. The experimental results demonstrate that our proposed FireFormer achieves state-of-the-art performance on the publicly available forest fire dataset FLAME—specifically, with an impressive 73.13% IoU and 84.48% F1 Score.

Assessment of the effectiveness of coarse resolution fire products in monitoring long-term changes in fire regime within protected areas in South Africa

Although high-resolution satellite and in situ airborne observations are becoming a common data source for fire monitoring, reconstruction of the historical fire record is based on less suitable data. In many regions, managers and policy-makers rely solely on coarse resolution global fire datasets. This study assessed the suitability of readily available products to accurately depict several fire metrics by comparing them to the manually derived fire scars within three natural reserves in South Africa (2003–2020). Contrary to previous findings, we showed that MCD64A1 and FireCCI51 products detected the majority of the burned area (70–97 %) within protected areas and showed high temporal consistency. Additionally, active and burned area products accurately detected the length and the peak of the fire season. Nonetheless, we found neither medium nor coarse resolution data suitable for evaluating long-term changes in fire size distribution. Coarse resolution burned area products demonstrated a very limited capacity to detect fires smaller than 100 ha, while a lack of sufficient non-cloudy Landsat images led to substantial gaps in fire reconstruction, bringing an unknown level of uncertainties in fire metrics estimates. Similarly, the only available medium-resolution burned area product (GABAM) was unsuitable for fire monitoring since unmapped areas were classified as unburned, making no distinction between regions with an absence of data and an absence of fires. Meanwhile, the performance of FireCCILT11 was ranked the lowest among all the available fire products. Given the lack of resources in many protected areas in South Africa and the pressing need to evaluate current and historical fire management practices and policies, we conclude that, in the absence of a more accurate higher-resolution alternative, MCD64A1 was the most suitable product for capturing year-to-year fluctuations in fire activity and a combination of MODIS and VIIRS active fire products for monitoring changes in fire seasonality.

YOLOv8-EMSC: A lightweight fire recognition algorithm for large spaces

Stringent fire prevention requirements are imperative in expansive environments. Fire detection in diverse large-scale settings typically relies on sensor-based or AI-driven target detection methods. Traditional fire detectors often suffer from false alarms and missed detections, failing to meet the fire safety requirements of large-scale structures. Many existing target detection algorithms are characterized by substantial model sizes. Some detection terminals in large structures face challenges deploying these models due to constrained computational resources. To address this issue, we propose a lightweight model, YOLOv8-EMSC, derived from YOLOv8n. The incorporation of C2f_EMSC, replacing the C2f module, significantly reduces the model parameters in the enhanced YOLOv8-EMSC model compared to YOLOv8n, thereby enhancing model inference speed. Extensive testing and validation using a custom-built large-scale infrared fire dataset demonstrates a 9.6 % reduction in parameters compared to the baseline model for YOLOv8-EMSC, achieving an average precision of 95.6 %, surpassing both the baseline and mainstream models and significantly enhancing fire detection accuracy in expansive environments.

Forty-Year Fire History Reconstruction from Landsat Data in Mediterranean Ecosystems of Algeria following International Standards

Algeria, the main fire hotspot on the southern rim of the Mediterranean Basin, lacks a complete fire dataset with official fire perimeters, and the existing one contains inconsistencies. Preprocessed global and regional burned area (BA) products provide valuable insights into fire patterns, characteristics, and dynamics over time and space, and into their impact on climate change. Nevertheless, they exhibit certain limitations linked with their inherent spatio-temporal resolutions as well as temporal and geographical coverage. To address the need for reliable BA information in Algeria, we systematically reconstructed, validated, and analyzed a 40-year (1984–2023) BA product (NEALGEBA; North Eastern ALGeria Burned Area) at 30 m spatial resolution in the typical Mediterranean ecosystems of this region, following international standards. We used Landsat data and the BA Mapping Tools (BAMTs) in the Google Earth Engine (GEE) to map BAs. The spatial validation of NEALGEBA, performed for 2017 and 2021 using independent 10 m spatial resolution Sentinel-2 reference data, showed overall accuracies > 98.10%; commission and omission errors < 8.20%; Dice coefficients > 91.90%; and relative biases < 3.44%. The temporal validation, however, using MODIS and VIIRS active fire hotspots, emphasized the limitation of Landsat-based BA products in temporal fire reporting accuracy terms. The intercomparison with five readily available BA products for 2017, by using the same validation process, demonstrated the overall outperformance of NEALGEBA. Furthermore, our BA product exhibited the highest correspondence with the ground-based BA estimates. NEALGEBA currently represents the most continuous and reliable time series of BA history at fine spatial resolution for NE Algeria, offering a significant contribution to further national and international fire hazard and impact assessments and acts as a reference dataset for contextualizing future weather extremes, such as the 2023 exceptional heat wave, which we show not to have led to the most extreme fire year over the last four decades.

Satellite Remote Sensing False Forest Fire Hotspot Excavating Based on Time-Series Features

Satellite remote sensing has become an important means of forest fire monitoring because it has the advantages of wide coverage, few ground constraints and high dynamics. When utilizing satellites for forest fire hotspot monitoring, two types of ground hotspots, agricultural and other fire hotspots can be ruled out through ground object features. False forest fire hotspots within forested areas must be excluded for a more accurate distinction between forest fires and non-forest fires. This study utilizes spatio-temporal data along with time-series classification to excavate false forest fire hotspots exhibiting temporal characteristics within forested areas and construct a dataset of such false forest fire hotspots, thereby achieving a more realistic forest fire dataset. Taking Hunan Province as the research object, this study takes the satellite ground hotspots in the forests of Hunan Province as the suspected forest fire hotspot dataset and excludes the satellite ground hotspots in the forests such as fixed heat sources, periodic heat sources and recurring heat sources which are excavated. The validity of these methods and results was then analyzed. False forest fire hotspots, from satellite ground hotspots extracted from 2019 to 2023 Himawari-8/9 satellite images, closely resemble the official release of actual forest fires data and the accuracy rate in the actual forest fire monitoring is 95.12%. This validates that the method employed in this study can improve the accuracy of satellite-based forest fire monitoring.

Satellite Image Cloud Automatic Annotator with Uncertainty Estimation

In satellite imagery, clouds obstruct the ground information, directly impacting various downstream applications. Thus, cloud annotation/cloud detection serves as the initial preprocessing step in remote sensing image analysis. Recently, deep learning methods have significantly improved in the field of cloud detection, but training these methods necessitates abundant annotated data, which requires experts with professional domain knowledge. Moreover, the influx of remote sensing data from new satellites has further led to an increase in the cost of cloud annotation. To address the dependence on labeled datasets and professional domain knowledge, this paper proposes an automatic cloud annotation method for satellite remote sensing images, CloudAUE. Unlike traditional approaches, CloudAUE does not rely on labeled training datasets and can be operated by users without domain expertise. To handle the irregular shapes of clouds, CloudAUE firstly employs a convex hull algorithm for selecting cloud and non-cloud regions by polygons. When selecting convex hulls, the cloud region is first selected, and points at the edges of the cloud region are sequentially selected as polygon vertices to form a polygon that includes the cloud region. Then, the same selection is performed on non-cloud regions. Subsequently, the fast KD-Tree algorithm is used for pixel classification. Finally, an uncertainty method is proposed to evaluate the quality of annotation. When the confidence value of the image exceeds a preset threshold, the annotation process terminates and achieves satisfactory results. When the value falls below the threshold, the image needs to undergo a subsequent round of annotation. Through experiments on two labeled datasets, HRC and Landsat 8, CloudAUE demonstrates comparable or superior accuracy to deep learning algorithms, and requires only one to two annotations to obtain ideal results. An unlabeled self-built Google Earth dataset is utilized to validate the effectiveness and generalizability of CloudAUE. To show the extension capabilities in various fields, CloudAUE also achieves desirable results on a forest fire dataset. Finally, some suggestions are provided to improve annotation performance and reduce the number of annotations.

Infrared/Visible Light Fire Image Fusion Method Based on Generative Adversarial Network of Wavelet-Guided Pooling Vision Transformer

To address issues of detail loss, limited matching datasets, and low fusion accuracy in infrared/visible light fire image fusion, a novel method based on the Generative Adversarial Network of Wavelet-Guided Pooling Vision Transformer (VTW-GAN) is proposed. The algorithm employs a generator and discriminator network architecture, integrating the efficient global representation capability of Transformers with wavelet-guided pooling for extracting finer-grained features and reconstructing higher-quality fusion images. To overcome the shortage of image data, transfer learning is utilized to apply the well-trained model to fire image fusion, thereby improving fusion precision. The experimental results demonstrate that VTW-GAN outperforms the DenseFuse, IFCNN, U2Fusion, SwinFusion, and TGFuse methods in both objective and subjective aspects. Specifically, on the KAIST dataset, the fusion images show significant improvements in Entropy (EN), Mutual Information (MI), and Quality Assessment based on Gradient-based Fusion (Qabf) by 2.78%, 11.89%, and 10.45%, respectively, over the next-best values. On the Corsican Fire dataset, compared to data-limited fusion models, the transfer-learned fusion images enhance the Standard Deviation (SD) and MI by 10.69% and 11.73%, respectively, and compared to other methods, they perform well in Average Gradient (AG), SD, and MI, improving them by 3.43%, 4.84%, and 4.21%, respectively, from the next-best values. Compared with DenseFuse, the operation efficiency is improved by 78.3%. The method achieves favorable subjective image outcomes and is effective for fire-detection applications.

A Machine Learning Approach to Predict the Occurrence of Forest Fires with Meteorological Parameters

Forest fires have emerged as a considerable challenge in the United States, posing substantial societal, economic, and environmental risks. As a result, the early and accurate prediction of these fires is imperative for management efforts. In this study, we used two Kaggle datasets: the “Algerian Forest Fires Dataset” with fire readings from 2012 and the “Forest Fires Data Set” with readings from 2007. However, because the second data set was originally intended for a regression task, providing approximate area values representing the predicted burned area of the forest fire, we phased the data set out while developing our final model. Ultimately, we used the Algerian Forest Fires Dataset, containing 13 attributes and 244 instances of forest fires in two regions of Algeria. To streamline the analysis, we reduced the number of features to 5, namely, month, temperature, humidity, wind, and rain. Moreover, we developed a Random Forest Classifier model to predict the occurrence of a forest fire, using the data set for training and testing. Performance was compared against Decision Tree, Logistic Regression, and Artificial Neural Network models, using cross-validation. The experiment showed a slight superiority to the Random Forest Classifier approach, achieving an accuracy score of 86.486% and an F1 score of 88.889%. Our approach provides a decimal value representing the probability for fire likelihood. Overarchingly, this research contributes to the advancement of forest fire prediction technologies by leveraging meteorological data.

DCP-Net: An Efficient Image Segmentation Model for Forest Wildfires

Wildfires usually lead to a large amount of property damage and threaten life safety. Image recognition for fire detection is now an important tool for intelligent fire protection, and the advancement of deep learning technologies has enabled an increasing number of cameras to possess functionalities for fire detection and automatic alarm triggering. To address the inaccuracies in extracting texture and positional information during intelligent fire recognition, we have developed a novel network called DCP-Net based on UNet, which excels at capturing flame features across multiple scales. We conducted experiments using the Corsican Fire Dataset produced by the “Environmental Science UMR CNRS 6134 SPE” laboratory at the University of Corsica and the BoWFire Dataset by Chino et al. Our algorithm was compared with networks such as SegNet, UNet, UNet++, and PSPNet, demonstrating superior performance across three metrics: mIoU, F1-score, and OA. Our proposed deep learning model achieves the best mIoU (78.9%), F1-score (76.1%), and OA (96.7%). These results underscore the robustness of our algorithm, which accurately identifies complex flames, thereby making a significant contribution to intelligent fire recognition. Therefore, the proposed DCP-Net model offers a viable solution to the challenges of wildfire monitoring using cameras, with hardware and software requirements typical of deep learning setups.

An open flame and smoke detection dataset for deep learning in remote sensing based fire detection

ABSTRACT The early warning of fires is pivotal for preventing substantial economic losses and ecological damage. However, the enhancement of fire model generalization performance still faces challenges due to limitations in existing fire datasets, such as image quantity and heterogeneity. Taking the advantage of rapid advancements in remote sensing technology and artificial intelligence, we introduce the Flame And Smoke Detection Dataset (FASDD), a pioneering collection comprising over 120,000 heterogeneous images covering various fire scenarios, aiming to promote the evolution of fire detection models. FASDD serves as a challenging benchmark for object detection tasks and features three sub-datasets: FASDD_CV, FASDD_UAV, and FASDD_RS. These sub-datasets comprise cross-domain images from ground-based, airborne, and spaceborne sensors, respectively. Extensive experiments using advanced Swin Transformer models demonstrate satisfactory fire detection performance, with mAP scores of 84.9%, 89.7%, and 74.0% on the respective sub-datasets. Notably, the application case in wildfire localization highlights the proficiency of FASDD_RS models for monitoring and tracking forest fires. Trained on FASDD_RS, FASDD_UAV, and FASDD_CV, deep learning models can be individually deployed on edge devices including satellites, drones, and terrestrial sensors, thus empowering collaborative fire warning within the space-air-ground integrated observation network. Additionally, a standardized procedure for data preparation, data procedure, and quality assurance is established, providing off-the-shelf annotations with four common configurations to facilitate machine learning model training. FASDD stands as a pivotal benchmark, providing valuable resources for training and evaluating deep learning fire detection models, especially Large Vision Models (LVM). It holds the potential to accelerate research progress in urban firefighting or forest fire detection, providing vital security measures for early fire warning and emergency response tasks.

A modified vision transformer architecture with scratch learning capabilities for effective fire detection

Fire is considered to be one of the major influencing factors that cause fatalities, property damage, and economic, and ecological disruption. To perform the early detection of fire using data from vision sensors and prevent or reduce the damage that fires cause, deep learning models have been widely adopted to overcome the limitations of conventional methods. However, mainstream convolutional neural network (CNN) models have limited generalisation abilities in unseen scenarios and struggle to obtain a good trade-off among the accuracy, inference speed, and model size. Currently, vision transformers (ViT) outperform conventional CNN models; however, they are computationally expensive and required more data for training. They provide a limited performance for small and medium-sized datasets, which are very common in the fire scene classification domain. In this work, we employ a novel ViT architecture by combining shifted patch tokenisation and local self-attention modules for efficient fire scene classification and enable the model to learn from scratch even on small and medium-sized datasets. Furthermore, to make the model suitable for real-time inferencing, we modify the transformer encoder and eventually achieve a reduced number of floating-point operations and a reduced model size. Additionally, in this work, a medium-scale fire dataset is developed that contains complex real-world scenarios. Our model is assessed on three benchmark and a self-created datasets using several evaluation metrics, including a novel cross-corpse evaluation metric, as well as a robustness evaluation metric. The experimental results indicate that our model achieved an overwhelmingly better performance compared to existing methods in terms of the accuracy and model complexity.