ESFD-YOLOv8n: Early Smoke and Fire Detection Method Based on an Improved YOLOv8n Model

Fire, Journal Year: 2024, Volume and Issue: 7(9), P. 303 - 303

Published: Aug. 27, 2024

Ensuring fire safety is essential to protect life and property, but modern infrastructure complex settings require advanced detection methods. Traditional object systems, often reliant on manual feature extraction, may fall short, while deep learning approaches are powerful, they can be computationally intensive, especially for real-time applications. This paper proposes a novel smoke method based the YOLOv8n model with several key architectural modifications. The standard Complete-IoU (CIoU) box loss function replaced more robust Wise-IoU version 3 (WIoUv3), enhancing predictions through its attention mechanism dynamic focusing. streamlined by replacing C2f module residual block, enabling targeted accelerating training inference, reducing overfitting. Integrating generalized efficient layer aggregation network (GELAN) blocks modules in neck of further enhances detection, optimizing gradient paths high performance. Transfer also applied enhance robustness. Experiments confirmed excellent performance ESFD-YOLOv8n, outperforming original 2%, 2.3%, 2.7%, mean average precision ([email protected]) 79.4%, 80.1%, recall 72.7%. Despite increased complexity, outperforms state-of-the-art algorithms meets requirements detection.

Language: Английский

---

MAG-FSNet:A high-precision robust forest fire smoke detection model integrating local features and global information

Measurement, Journal Year: 2025, Volume and Issue: unknown, P. 116813 - 116813

Published: Jan. 1, 2025

Language: Английский

---

SIMCB-Yolo: An Efficient Multi-Scale Network for Detecting Forest Fire Smoke

Forests, Journal Year: 2024, Volume and Issue: 15(7), P. 1137 - 1137

Published: June 29, 2024

Forest fire monitoring plays a crucial role in preventing and mitigating forest disasters. Early detection of smoke is essential for timely response to emergencies. The key effective lies accounting the various levels targets images, enhancing model’s anti-interference capabilities against mountain clouds fog, reducing false positives missed detections. In this paper, we propose an improved multi-level model based on You Only Look Once v5s (Yolov5s) called SIMCB-Yolo. This aims achieve high-precision at levels. First, address issue low precision detecting small target smoke, Swin transformer head added neck Yolov5s, detection. Then, detections due decline conventional accuracy after improving accuracy, introduced cross stage partial network bottleneck with three convolutional layers (C3) channel block sequence (CBS) into trunk. These additions help extract more surface features enhance smoke. Finally, SimAM attention mechanism complex background interference detection, further Experimental results demonstrate that, compared Yolov5s model, SIMCB-Yolo achieves average recognition (mAP50) 85.6%, increase 4.5%. Additionally, mAP50-95 63.6%, improvement 6.9%, indicating good accuracy. performance self-built dataset also significantly better than that current mainstream models, demonstrating high practical value.

Language: Английский

---

A Lightweight Dynamically Enhanced Network for Wildfire Smoke Detection in Transmission Line Channels

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

---

FACNet: A high-precision pumpkin seedling point cloud organ segmentation method

Computers and Electronics in Agriculture, Journal Year: 2025, Volume and Issue: 231, P. 110049 - 110049

Published: Feb. 1, 2025

Language: Английский

---

Deep Residual Multi-resolution Features and Optimized Kernel ELM for Forest Fire Image Detection Using Imbalanced Database

Fire Technology, Journal Year: 2025, Volume and Issue: unknown

Published: March 31, 2025

Language: Английский

---

Multi-level Graph Subspace Contrastive Learning for Hyperspectral Image Clustering

2022 International Joint Conference on Neural Networks (IJCNN), Journal Year: 2024, Volume and Issue: 36, P. 1 - 8

Published: June 30, 2024

Language: Английский

---

Improved Lightweight YOLOv11 Algorithm for Real-Time Forest Fire Detection

Electronics, Journal Year: 2025, Volume and Issue: 14(8), P. 1508 - 1508

Published: April 9, 2025

Modern computer vision techniques for forest fire detection face a trade-off between computational efficiency and accuracy in complex environments. To address this, we propose lightweight YOLOv11n-based framework optimized edge deployment. The backbone network integrates novel C3k2MBNV2 (Cross Stage Partial Bottleneck with 3 convolutions kernel size 2 MobileNetV2) block to enable efficient feature extraction via compact architecture. We further introduce the SCDown (Spatial-Channel Decoupled Downsampling) both neck preserve critical information during downsampling. incorporates C3k2WTDC 2, combined Wavelet Transform Depthwise Convolution) block, enhancing contextual understanding reduced overhead. Experiments on dataset demonstrate that our model achieves 53.2% reduction parameters 28.6% fewer FLOPs compared YOLOv11n (You Only Look Once version eleven), along 3.3% improvement mean average precision. These advancements establish an optimal balance accuracy, enabling proposed attain real-time capabilities resource-constrained devices This work provides practical solution deploying reliable systems scenarios demanding low latency minimal resources.

Language: Английский

---

YOLO-SAD for fire detection and localization in real-world images

Digital Signal Processing, Journal Year: 2025, Volume and Issue: unknown, P. 105320 - 105320

Published: May 1, 2025

Language: Английский

---

SDRnet: A Deep Fusion Network for ISAR Ship Target Recognition Based on Feature Separation and Weighted Decision

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(11), P. 1920 - 1920

Published: May 27, 2024

Existing methods for inverse synthetic aperture radar (ISAR) target recognition typically rely on a single high-resolution signal type, such as ISAR images or range profiles (HRRPs). However, and HRRP data offer representations of targets across different aspects, each containing valuable information crucial recognition. Moreover, the process generating inherently facilitates acquisition data, ensuring timely collection. Therefore, to fully leverage from both enhance ship performance, we propose novel deep fusion network named Separation-Decision Recognition (SDRnet). First, our approach employs convolutional neural (CNN) extract initial feature vectors data. Subsequently, separation module is employed derive more robust representation. Finally, introduce weighted decision overall predictive performance. We validate method using simulated measured ten categories targets. The experimental results confirm effectiveness in improving

Language: Английский

---