Published: Dec. 27, 2024
Language: Английский
Fire, Journal Year: 2025, Volume and Issue: 8(2), P. 59 - 59
Published: Jan. 30, 2025
Forest fires pose a severe threat to ecological environments and the safety of human lives property, making real-time forest fire monitoring crucial. This study addresses challenges in image object detection, including small targets, sparse smoke, difficulties feature extraction, by proposing TFNet, Transformer-based multi-scale fusion detection network. TFNet integrates several components: SRModule, CG-MSFF Encoder, Decoder Head, WIOU Loss. The SRModule employs multi-branch structure learn diverse representations images, utilizing 1 × convolutions generate redundant maps enhance diversity. Encoder introduces context-guided attention mechanism combined with adaptive (AFF), enabling effective reweighting features across layers extracting both local global representations. Head refine output iteratively optimizing target queries using self- cross-attention, improving accuracy. Additionally, Loss assigns varying weights IoU metric for predicted versus ground truth boxes, thereby balancing positive negative samples localization Experimental results on two publicly available datasets, D-Fire M4SFWD, demonstrate that outperforms comparative models terms precision, recall, F1-Score, mAP50, mAP50–95. Specifically, dataset, achieved metrics 81.6% 74.8% an F1-Score 78.1%, mAP50 81.2%, mAP50–95 46.8%. On M4SFWD these improved 86.6% 83.3% 84.9%, 89.2%, 52.2%. proposed offers technical support developing efficient practical systems.
Language: Английский
Citations
2
Fire, Journal Year: 2025, Volume and Issue: 8(1), P. 26 - 26
Published: Jan. 13, 2025
Forest fires cause extensive environmental damage, making early detection crucial for protecting both nature and communities. Advanced computer vision techniques can be used to detect smoke fire. However, accurate of fire in forests is challenging due different factors such as shapes, changing light, similarity with other smoke-like elements clouds. This study explores recent YOLO (You Only Look Once) deep-learning object models YOLOv9, YOLOv10, YOLOv11 detecting forest environments. The evaluation focuses on key performance metrics, including precision, recall, F1-score, mean average precision (mAP), utilizes two benchmark datasets featuring diverse instances across findings highlight the effectiveness small version (YOLOv9t, YOLOv10n, YOLOv11n) tasks. Among these, YOLOv11n demonstrated highest performance, achieving a 0.845, recall 0.801, mAP@50 0.859, mAP@50-95 0.558. versions (YOLOv11n YOLOv11x) were evaluated compared against several studies that employed same datasets. results show YOLOv11x delivers promising variants models.
Language: Английский
Citations
1
Measurement, Journal Year: 2025, Volume and Issue: unknown, P. 116813 - 116813
Published: Jan. 1, 2025
Language: Английский
Citations
1
Forests, Journal Year: 2025, Volume and Issue: 16(2), P. 201 - 201
Published: Jan. 22, 2025
Detecting wildfires and smoke is essential for safeguarding forest ecosystems offers critical information the early evaluation prevention of such incidents. The advancement unmanned aerial vehicle (UAV) remote sensing has further enhanced detection smoke, which enables rapid accurate identification. This paper presents an integrated one-stage object framework designed simultaneous identification in UAV imagery. By leveraging mixed data augmentation techniques, enriches dataset with small targets to enhance its performance targets. A novel backbone enhancement strategy, integrating region convolution feature refinement modules, developed facilitate ability localize features high transparency within complex backgrounds. shape aware loss function, proposed effective capture irregularly shaped fire edges, facilitating localization smoke. Experiments conducted on a demonstrate that achieves promising terms both accuracy speed. attains mean Average Precision (mAP) 79.28%, F1 score 76.14%, processing speed 8.98 frames per second (FPS). These results reflect increases 4.27%, 1.96%, 0.16 FPS compared YOLOv10 model. Ablation studies validate incorporation augmentation, models, substantial improvements over findings highlight framework’s capability rapidly effectively identify using imagery, thereby providing valuable foundation proactive measures.
Language: Английский
Citations
0
Processes, Journal Year: 2025, Volume and Issue: 13(2), P. 349 - 349
Published: Jan. 27, 2025
In view of the problems that mean existing detection networks are not effective in detecting dynamic targets such as wildfire smoke, a lightweight dynamically enhanced transmission line channel smoke network LDENet is proposed. Firstly, Dynamic Lightweight Conv Module (DLCM) devised within backbone YOLOv8 to enhance perception flames and through convolution. Then, Ghost used model. DLCM reduces number model parameters improves accuracy detection. DySample upsampling operator part make image generation more accurate with very few parameters. Finally, course training process, loss function improved. EMASlideLoss improve ability for small targets, Shape-IoU optimize shape wildfires smoke. Experiments conducted on datasets, final mAP50 86.6%, which 1.5% higher than YOLOv8, decreased by 29.7%. The experimental findings demonstrate capable effectively ensuring safety corridors.
Language: Английский
Citations
0
Fire, Journal Year: 2025, Volume and Issue: 8(2), P. 67 - 67
Published: Feb. 7, 2025
An aircraft hangar is a special large-space environment containing lot of combustible materials and high-value equipment. It essential to quickly accurately detect early-stage fires when they occur. In this study, experiments were conducted in real simulate the occurrence fires, collected images classified, labeled, organized form dataset used paper. The fire data categorized into two target classes: smoke. This study proposes an detection method that integrates attention mechanism, which was based on You Only Look Once Version 8 Nano (YOLOv8n) framework further improved. Technically, optimization YOLOv8n mainly carried out stages: Firstly, at network structure level, neck reconstructed using large separable kernel (LSKA) module; secondly, terms loss function design, original CIoU replaced with dynamic focus-based Wise-IoU enhance performance model. new algorithm named LSKA-YOLOv8n+WIoU. Experimental results show LSKA-YOLOv8n+WIoU has superior compared related state-of-the-art algorithms. Compared model, precision increased by 10% 86.7%, recall 8.8% 67.2%, mean average (mAP) 5.9% 69.5%. parameter size reduced 0.5MB 5.7MB. Through these improvements, accuracy flame smoke enhanced while reducing computational complexity, increasing efficiency, effectively mitigating phenomena missed false detections. contributes enhancing speed systems environments, providing reliable support for alarm work.
Language: Английский
Citations
0
Fire, Journal Year: 2025, Volume and Issue: 8(4), P. 142 - 142
Published: April 2, 2025
The increasing frequency and severity of agricultural fires pose significant threats to food security, economic stability, environmental sustainability. Traditional fire-detection methods, relying on satellite imagery ground-based sensors, often suffer from delayed response times high false-positive rates, limiting their effectiveness in mitigating fire-related damages. In this study, we propose an advanced deep learning-based framework that integrates the Single-Shot MultiBox Detector (SSD) with computationally efficient MobileNetV2 architecture. This integration enhances real-time fire- smoke-detection capabilities while maintaining a lightweight deployable model suitable for Unmanned Aerial Vehicle (UAV)-based monitoring. proposed was trained evaluated custom dataset comprising diverse fire scenarios, including various conditions intensities. Comprehensive experiments comparative analyses against state-of-the-art object-detection models, such as You Only Look Once (YOLO), Faster Region-based Convolutional Neural Network (Faster R-CNN), SSD-based variants, demonstrated superior performance our model. results indicate approach achieves mean Average Precision (mAP) 97.7%, significantly surpassing conventional models detection speed 45 frames per second (fps) requiring only 5.0 GFLOPs computational power. These characteristics make it particularly deployment edge-computing environments, UAVs remote monitoring systems.
Language: Английский
Citations
0
Fire, Journal Year: 2025, Volume and Issue: 8(5), P. 165 - 165
Published: April 22, 2025
Fires pose significant threats to human safety, health, and property. Traditional methods, with their inefficient use of features, struggle meet the demands fire detection. You Only Look Once (YOLO), as an efficient deep learning object detection framework, can rapidly locate identify smoke objects in visual images. However, research utilizing latest YOLO11 for remains sparse, addressing scale variability well practicality models continues be a focus. This study first compares classic YOLO series analyze its advantages tasks. Then, tackle challenges model practicality, we propose Multi-Scale Convolutional Attention (MSCA) mechanism, integrating it into create YOLO11s-MSCA. Experimental results show that outperforms other by balancing accuracy, speed, practicality. The YOLO11s-MSCA performs exceptionally on D-Fire dataset, improving overall accuracy 2.6% recognition 2.8%. demonstrates stronger ability small objects. Although remain handling occluded targets complex backgrounds, exhibits strong robustness generalization capabilities, maintaining performance complicated environments.
Language: Английский
Citations
0
Forests, Journal Year: 2024, Volume and Issue: 15(10), P. 1781 - 1781
Published: Oct. 10, 2024
Forest fires pose a significant threat to forest resources and wildlife. To balance accuracy parameter efficiency in fire detection, this study proposes an improved model, Mcan-YOLO, based on YOLOv7. In the Neck section, asymptotic feature pyramid network (AFPN) was employed effectively capture multi-scale information, replacing traditional module. Additionally, content-aware reassembly of features (CARAFE) replaced conventional upsampling method, further reducing number parameters. The normalization-based attention module (NAM) integrated after ELAN-T enhance recognition various smoke features, Mish activation function used optimize model convergence. A real dataset constructed using mean structural similarity (MSSIM) algorithm for training validation. experimental results showed that, compared YOLOv7-tiny, Mcan-YOLO precision by 4.6%, recall 6.5%, mAP50 4.7%, while parameters 5%. Compared with other mainstream algorithms, achieved better fewer
Language: Английский
Citations
1
Fire, Journal Year: 2024, Volume and Issue: 7(12), P. 428 - 428
Published: Nov. 22, 2024
Wildfires and drought stressors can significantly limit forest recovery in Mediterranean-type ecosystems. Since 2010, the region of central Chile has experienced a prolonged Mega Drought, which intensified into Hyper Drought 2019, characterized by record-low precipitation high temperatures, further constraining recovery. This study evaluates short-term (5-year) post-fire vegetation across gradients two types evergreen sclerophyllous forests thorny shrubland, analyzing Landsat time series (1987–2022) from 42 wildfires. Using LandTrendr algorithm, we assessed based on NDVI changes between pre-fire values subsequent years. The results reveal significant differences during period, among three studied. xeric forest, dominated Quillaja saponaria Lithrea caustica, showed interaction effects levels fire severity, while shrubland displayed no effects. mesic Cryptocarya alba Peumus boldus, exhibited additional periods, along with These findings underscore critical role prolonged, severe shaping dynamics highlight need to understand these patterns improve future resilience under increasingly arid conditions.
Language: Английский
Citations
0