A Comparative Analysis of YOLOv9, YOLOv10, YOLOv11 for Smoke and Fire Detection

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

---

Hyperparameter optimization of YOLOv8 for smoke and wildfire detection: Implications for agricultural and environmental safety

Artificial Intelligence in Agriculture, Journal Year: 2024, Volume and Issue: 12, P. 109 - 126

Published: May 31, 2024

In this study, we extensively evaluated the viability of state-of-the-art YOLOv8 architecture for object detection tasks, specifically tailored smoke and wildfire identification with a focus on agricultural environmental safety. All available versions were initially fine-tuned domain-specific dataset that included variety scenarios, crucial comprehensive monitoring. The 'large' version (YOLOv8l) was selected further hyperparameter tuning based its performance metrics. This model underwent detailed optimization using One Factor At Time (OFAT) methodology, concentrating key parameters such as learning rate, batch size, weight decay, epochs, optimizer. Insights from OFAT study used to define search spaces subsequent Random Search (RS). final derived RS demonstrated significant improvements over initial model, increasing overall precision by 1.39 %, recall 1.48 F1-score 1.44 [email protected] 0.70 protected]:0.95 5.09 %. We validated enhanced model's efficacy diverse set real-world images, reflecting various settings, confirm robustness in detecting fire. These results underscore reliability effectiveness scenarios critical safety work, representing advancement field fire through machine learning, lays strong foundation future research solutions aimed at safeguarding areas natural environments.

Language: Английский

---

A comparative study of YOLOv5 and YOLOv8 for corrosion segmentation tasks in metal surfaces

Array, Journal Year: 2024, Volume and Issue: 22, P. 100351 - 100351

Published: June 1, 2024

This study delves into the comparative efficacy of YOLOv5 and YOLOv8 in corrosion segmentation tasks. We employed three unique datasets, comprising 4942, 5501, 6136 images, aiming to thoroughly evaluate models' adaptability robustness diverse scenarios. The assessment metrics included precision, recall, F1-score, mean average precision. Furthermore, graphical tests offered a visual perspective on capabilities each architecture. Our results highlight YOLOv8's superior speed accuracy across further corroborated by evaluations. These assessments were instrumental emphasizing proficiency handling complex corroded surfaces. However, largest dataset, both models encountered challenges, particularly with overlapping bounding boxes. notably lagged, struggling achieve performance standards set YOLOv8, especially irregular In conclusion, our findings underscore enhanced capabilities, establishing it as preferable choice for real-world detection research thus offers invaluable insights, poised redefine management strategies guide future explorations identification.

Language: Английский

---

Computer vision for wildfire detection: a critical brief review

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: 83(35), P. 83427 - 83470

Published: March 13, 2024

Language: Английский

---

A Study of YOLO Architectures for Wildfire and Smoke Detection in Ground and Aerial Imagery

Results in Engineering, Journal Year: 2025, Volume and Issue: unknown, P. 104869 - 104869

Published: April 1, 2025

Language: Английский

---

An End-to-End Platform for Managing Third-Party Risks in Oil Pipelines

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 77831 - 77851

Published: Jan. 1, 2024

Ensuring the safe and reliable operation of underground oil pipelines is crucial to prevent environmental disasters maintain uninterrupted energy supply. Yet, this vast network faces threats from third-party activities, natural disasters, aging infrastructure, posing risks catastrophic consequences if left unaddressed. In response need, paper presents a computer vision system for detecting (vehicular movement) near pipelines. Our primary objective showcase practical application cutting-edge models in real-world operational environments. For this, we construct dataset comprising 1,003 aerial images, covering seven classes vehicles commonly encountered pipelines, including trucks, forklifts, machinery, pickups, tractors, vehicles, buses. This serves as foundation training hyperparameter optimization YOLOv8x-based detection model, used work. The optimized model exhibits strong performance across precision, recall, F1-score, mean average precision metrics compared baseline model. Additionally, graphical tests illustrated that demonstrates higher confidence scores reduction false positives. addition, platform has been developed seamlessly integrate offers range functionalities, enabling users access alert history, prioritize alerts, track actions taken on each alert, visualize alerts geographically, receive notifications identified risks, generate detailed reports comprehensive analysis decision-making.

Language: Английский

---

A deep learning object detection method for fracture identification using conventional well logs

IEEE Transactions on Geoscience and Remote Sensing, Journal Year: 2024, Volume and Issue: 62, P. 1 - 16

Published: Jan. 1, 2024

Language: Английский

---

Automatic Vehicle Accident Detection and Classification from Images: A Comparison of YOLOv9 and YOLO-NAS Algorithms

Published: May 15, 2024

Language: Английский

---

EngineFaultDB: A Novel Dataset for Automotive Engine Fault Classification and Baseline Results

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 126155 - 126171

Published: Jan. 1, 2023

This paper introduces EngineFaultDB, a novel dataset capturing the intricacies of automotive engine diagnostics. Centered around widely represented C14NE spark ignition engine, data was collected under controlled laboratory conditions, simulating various operational states, including normal and specific fault scenarios. Utilizing tools such as an NGA 6000 gas analyzer USB 6008 acquisition card from National Instruments, we were able to monitor capture comprehensive range parameters, throttle position fuel consumption exhaust emissions. Our dataset, comprising 55,999 meticulously curated entries across 14 distinct variables, provides holistic picture behavior, making it invaluable resource for researchers practitioners. For evaluation, several classifiers, logistic regression, decision trees, random forests, support vector machines, k-nearest neighbors, feed-forward neural network, trained on this dataset. Their performance, standard configurations simple network architecture, offers foundational benchmarks future explorations. Results underscore dataset's potential in fostering advanced diagnostic algorithms. As testament our commitment open research, EngineFaultDB is freely available academic use. Future work involves expanding diversity, exploring deeper architectures, integrating real-world conditions.

Language: Английский

---

VHRTrees: a new benchmark dataset for tree detection in satellite imagery and performance evaluation with YOLO-based models

Frontiers in Forests and Global Change, Journal Year: 2025, Volume and Issue: 7

Published: Jan. 6, 2025

Natural and planted forests, covering approximately 31% of the Earth’s land area, are crucial for global ecosystems, providing essential services such as regulating water cycle, soil conservation, carbon storage, biodiversity preservation. However, traditional forest mapping monitoring methods often costly limited in scale, highlighting need to develop innovative approaches tree detection that can enhance management. In this study, we present a new dataset detection, VHRTrees, derived from very high-resolution RGB satellite images. This includes 26,000 boundaries 1,496 image patches different geographical regions, representing various topographic climatic conditions. We implemented object algorithms evaluate performance methods, propose best experimental configurations, generate benchmark analysis further studies. conducted our experiments with variants hyperparameter settings YOLOv5, YOLOv7, YOLOv8, YOLOv9 models. Results extensive indicate that, increasing network resolution batch size led higher precision recall detection. YOLOv8m, optimized Auto, achieved highest F1-score (0.932) mean Average Precision (mAP)@0.50 Intersection over Union threshold (0.934), although some other configurations showed [email protected]:0.95. These findings underscore effectiveness You Only Look Once (YOLO)-based real-time applications, offering cost-effective accurate solution using imagery. The VHRTrees dataset, related source codes, pretrained models available at https://github.com/RSandAI/VHRTrees .

Language: Английский

---