Video Fire Detection Methods Based on Deep Learning: Datasets, Methods, and Future Directions

Fire, Год журнала: 2023, Номер 6(8), С. 315 - 315

Опубликована: Авг. 14, 2023

Among various calamities, conflagrations stand out as one of the most-prevalent and -menacing adversities, posing significant perils to public safety societal progress. Traditional fire-detection systems primarily rely on sensor-based detection techniques, which have inherent limitations in accurately promptly detecting fires, especially complex environments. In recent years, with advancement computer vision technology, video-oriented fire owing their non-contact sensing, adaptability diverse environments, comprehensive information acquisition, progressively emerged a novel solution. However, approaches based handcrafted feature extraction struggle cope variations smoke or flame caused by different combustibles, lighting conditions, other factors. As powerful flexible machine learning framework, deep has demonstrated advantages video detection. This paper summarizes deep-learning-based video-fire-detection methods, focusing advances commonly used datasets for recognition, object detection, segmentation. Furthermore, this provides review outlook development prospects field.

Язык: Английский

---

Forest fire size amplifies postfire land surface warming

Nature, Год журнала: 2024, Номер 633(8031), С. 828 - 834

Опубликована: Сен. 25, 2024

Climate warming has caused a widespread increase in extreme fire weather, making forest fires longer-lived and larger

Язык: Английский

---

Advancements in remote sensing for active fire detection: A review of datasets and methods

The Science of The Total Environment, Год журнала: 2024, Номер 943, С. 173273 - 173273

Опубликована: Май 31, 2024

Язык: Английский

---

LUFFD-YOLO: A Lightweight Model for UAV Remote Sensing Forest Fire Detection Based on Attention Mechanism and Multi-Level Feature Fusion

Remote Sensing, Год журнала: 2024, Номер 16(12), С. 2177 - 2177

Опубликована: Июнь 15, 2024

The timely and precise detection of forest fires is critical for halting the spread wildfires minimizing ecological economic damage. However, large variation in target size complexity background UAV remote sensing images increase difficulty real-time fire detection. To address this challenge, study proposes a lightweight YOLO model (LUFFD-YOLO) based on attention mechanism multi-level feature fusion techniques: (1) GhostNetV2 was employed to enhance conventional convolution YOLOv8n decreasing number parameters model; (2) plug-and-play enhanced small-object C2f (ESDC2f) structure proposed capability small fires; (3) an innovative hierarchical feature-integrated (HFIC2f) improve model’s ability extract information from complex backgrounds fusion. LUFFD-YOLO surpasses YOLOv8n, achieving 5.1% enhancement mAP 13% reduction parameter count obtaining desirable generalization different datasets, indicating good balance between high accuracy efficiency. This work would provide significant technical support using remote-sensing images.

Язык: Английский

---

Wintertime Heavy Haze Episodes in Northeast China Driven by Agricultural Fire Emissions

Environmental Science & Technology Letters, Год журнала: 2024, Номер 11(2), С. 150 - 157

Опубликована: Янв. 22, 2024

Heavy haze events occur frequently over northeast China during the winter, despite successful implementation of Clean Air Act, which primarily targets fossil fuel sources, in recent years. Agricultural fires have been suggested as one main causes these episodes. However, their regional contribution to fine particulate matter (PM2.5) pollution has not systematically evaluated. In this study, we use GEOS-Chem model investigate role agricultural heavy episodes from December 2018 March 2019 Heilongjiang province. Our results show significant discrepancies between simulated and observed PM2.5 concentrations severe days. By increasing fire emissions GFED4s inventory by a factor ∼23, are able better replicate model, indicating under-representation inventory. Furthermore, baseline simulation overestimates black carbon organic ratio Harbin, suggesting biased emission specified assessment underscores that agriculture constitute cause extreme study period, strictly implemented ban would improve air quality with substantial health benefits.

Язык: Английский

---

Remote sensing for wildfire monitoring: Insights into burned area, emissions, and fire dynamics

One Earth, Год журнала: 2024, Номер 7(6), С. 1022 - 1028

Опубликована: Июнь 1, 2024

Remote sensing plays a central role in monitoring wildfires throughout their life cycle, including assessing pre-fire fuel conditions, characterizing active fire locations and emissions, evaluating post-fire effects on vegetation, air quality, climate. This primer examines current remote products used wildfire research, focusing application deriving burned area emissions data tracking the dynamic spread of individual events. We evaluate strengths weaknesses these address key challenges such as generating complete, continuous, consistent long-term data. also explore future opportunities directions technology for characterization management.

Язык: Английский

---

Hourly biomass burning emissions product from blended geostationary and polar-orbiting satellites for air quality forecasting applications

Remote Sensing of Environment, Год журнала: 2022, Номер 281, С. 113237 - 113237

Опубликована: Сен. 6, 2022

Язык: Английский

---

A Forest Fire Susceptibility Modeling Approach Based on Light Gradient Boosting Machine Algorithm

Remote Sensing, Год журнала: 2022, Номер 14(17), С. 4362 - 4362

Опубликована: Сен. 2, 2022

A forest fire susceptibility map generated with the model is basis of prevention resource allocation. more reliable helps improve effectiveness Thus, further improving prediction accuracy always goal modeling. This paper developed a based on an ensemble learning method, namely light gradient boosting machine (LightGBM), to produce accurate map. In modeling, subtropical national park in Jiangsu province China was used as case study area. We collected and selected eight variables from occurrence driving factors for modeling correlation analysis. These are topographic factors, climatic human activity vegetation factors. For comparative analysis, another two popular methods, logistic regression (LR) random (RF) were also applied construct models. The results show that temperature main factor produced map, extremely high areas classified by LR, RF, LightGBM 5.82%, 18.61%, 19%, respectively. F1-score higher than LR RF LightGBM, 88.8%, 84.8%, 82.6%, area under curve (AUC) them 0.935, 0.918, 0.868, introduced method shows better ability performance evaluation metrics.

Язык: Английский

---

Global Warming Reshapes European Pyroregions

Earth s Future, Год журнала: 2023, Номер 11(5)

Опубликована: Май 1, 2023

Abstract Wildland fire is expected to increase in response global warming, yet little known about future changes regimes Europe. Here, we developed a pyrogeography based on statistical models better understand how warming reshapes across the continent. We identified five large‐scale pyroregions with different levels of area burned, frequency, intensity, length period, size distribution, and seasonality. All other things being equal, was found alter distribution these pyroregions, an expansion most prone ranging respectively from 50% 130% under 2° 4°C scenarios. Our estimates indicate strong amplification parts southern Europe subsequent shift toward new regimes, implying substantial socio‐ecological impacts absence mitigation or adaptation measures.

Язык: Английский

---

A self-adaptive wildfire detection algorithm by fusing physical and deep learning schemes

International Journal of Applied Earth Observation and Geoinformation, Год журнала: 2024, Номер 127, С. 103671 - 103671

Опубликована: Янв. 27, 2024

Currently, the spectra-based physical models and deep learning methods are frequently used to detect wildfires from remote sensing data. However, algorithms mainly rely on radiative transfer processes, which limit their effectiveness in detecting small weak fires. On other hand, usually lack mechanism constraints, thus generally resulting false alarms of bright surfaces. It is promising combine advantages them correspondingly reduce inherent error a single algorithm. To this end, paper, both local contextual global index method based mechanisms optimized, simultaneously, new U-Net model also establish accurately Moreover, YOLO v5 incorporated for first time extract remove objects with high exposure. Based above series novel works, self-adaptive fusing algorithm finally proposed. Our results reveal that: (1) Short-wave infrared band about 2.15 μm crucial fire detection data moderate-to-high resolutions. Taking Landsat 8 as an example, combinations 7, 6, 2(SWIR + VI), 5(SWIR NIR), 5, 3(SWIR VI NIR) show reasonable accuracy, recall rate greater than 81 %. The thermal can be assist general location serve alternative choice extreme cases. (2) optimized predict more accurate positions. (3) very effective introduce framework exposure urban suburban regions. (4) proposed fusion integrates various schemes, proving its better performance terms robustness, stability generality compared any method. Even situations such Gobi Desert, thin cloud edges, mountain shadow areas, still works well. tests Sentinel-2A, WorldView-3, SPOT-4 potential applicability newly algorithm, especially fine spatial spectral

Язык: Английский

---