TFNet: Transformer-Based Multi-Scale Feature Fusion Forest Fire Image Detection Network

Fire, Год журнала: 2025, Номер 8(2), С. 59 - 59

Опубликована: Янв. 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.

Язык: Английский

---

Forest fire probability zonation using dNBR and machine learning models: a case study at the Similipal Biosphere Reserve (SBR), Odisha, India

Environmental Science and Pollution Research, Год журнала: 2025, Номер unknown

Опубликована: Янв. 30, 2025

Язык: Английский

---

Multisensor Integrated Drought Severity Index (IDSI) for assessing agricultural drought in Odisha, India

Remote Sensing Applications Society and Environment, Год журнала: 2024, Номер 37, С. 101399 - 101399

Опубликована: Ноя. 13, 2024

Язык: Английский

---

Predictive machine learning and geospatial modeling reveal PM10 hotspots and guide targeted air pollution interventions in Addis Ababa, Ethiopia

Deleted Journal, Год журнала: 2025, Номер 7(4)

Опубликована: Март 26, 2025

Язык: Английский

---

Forecasting shoreline dynamics and land use/land cover changes in Balukhand-Konark Wildlife Sanctuary (India) using geospatial techniques and machine learning

The Science of The Total Environment, Год журнала: 2025, Номер 975, С. 179207 - 179207

Опубликована: Апрель 7, 2025

Язык: Английский

---

Climate variability and forest fires: trends, correlation, spatiotemporal patterns in the Seven Sister States of northeastern India (2001–2022)

Remote Sensing Applications Society and Environment, Год журнала: 2025, Номер unknown, С. 101586 - 101586

Опубликована: Май 1, 2025

Язык: Английский

---

Development of a novel Slope-Integrated Drought Index (SIDI) for comprehensive drought assessment

Theoretical and Applied Climatology, Год журнала: 2025, Номер 156(6)

Опубликована: Май 21, 2025

Язык: Английский

---

CN2VF-Net: A Hybrid Convolutional Neural Network and Vision Transformer Framework for Multi-Scale Fire Detection in Complex Environments

Fire, Год журнала: 2025, Номер 8(6), С. 211 - 211

Опубликована: Май 26, 2025

Fire detection remains a challenging task due to varying fire scales, occlusions, and complex environmental conditions. This paper proposes the CN2VF-Net model, novel hybrid architecture that combines vision Transformers (ViTs) convolutional neural networks (CNNs), effectively addressing these challenges. By leveraging global context understanding of ViTs local feature extraction capabilities CNNs, model learns multi-scale attention mechanism dynamically focuses on regions at different thereby improving accuracy robustness. The evaluation D-Fire dataset demonstrate proposed achieves mean average precision an IoU threshold 0.5 (mAP50) 76.1%, F1-score 81.5%, recall 82.8%, 83.3%, (mIoU50–95) 77.1%. These results outperform existing methods by 1.6% in precision, 0.3% recall, 3.4% F1-score. Furthermore, visualizations such as Grad-CAM heatmaps prediction overlays provide insight into model’s decision-making process, validating its capability detect segment regions. findings underscore effectiveness applicability real-world monitoring systems. With superior performance interpretability, sets new benchmark segmentation, offering reliable approach protecting life, property, environment.

Язык: Английский

---

Examining the drivers of forest cover change and deforestation susceptibility in Northeast India using multicriteria decision-making models

Environmental Monitoring and Assessment, Год журнала: 2024, Номер 196(11)

Опубликована: Окт. 24, 2024

Язык: Английский

---