A Forest Fire Detection System Based on Ensemble Learning

Xu, Runsheng; Lin, Haifeng; Lu, Kangjie; Cao, Lin; Liu, Yunfei

doi:10.3390/f12020217

Cited by 369 publications

(189 citation statements)

References 31 publications

Supporting

Mentioning

184

Contrasting

Unclassified

Order By: Relevance

“…At the same time, BiFPN also shows the best efficiency in multi-scale feature fusion. At present, the deep learning model based on EfficientDet and BiFPN is being applied to a variety of research fields, such as forest fire prevention (Xu et al, 2021), estimation of fashion landmarks (Kim et al, 2021), detection of garbage scattering areas (You et al, 2020), etc.…”

Section: Introductionmentioning

confidence: 99%

Occlusion Robust Wheat Ear Counting Algorithm Based on Deep Learning

2021

View full text Add to dashboard Cite

Counting the number of wheat ears in images under natural light is an important way to evaluate the crop yield, thus, it is of great significance to modern intelligent agriculture. However, the distribution of wheat ears is dense, so the occlusion and overlap problem appears in almost every wheat image. It is difficult for traditional image processing methods to solve occlusion problem due to the deficiency of high-level semantic features, while existing deep learning based counting methods did not solve the occlusion efficiently. This article proposes an improved EfficientDet-D0 object detection model for wheat ear counting, and focuses on solving occlusion. First, the transfer learning method is employed in the pre-training of the model backbone network to extract the high-level semantic features of wheat ears. Secondly, an image augmentation method Random-Cutout is proposed, in which some rectangles are selected and erased according to the number and size of the wheat ears in the images to simulate occlusion in real wheat images. Finally, convolutional block attention module (CBAM) is adopted into the EfficientDet-D0 model after the backbone, which makes the model refine the features, pay more attention to the wheat ears and suppress other useless background information. Extensive experiments are done by feeding the features to detection layer, showing that the counting accuracy of the improved EfficientDet-D0 model reaches 94%, which is about 2% higher than the original model, and false detection rate is 5.8%, which is the lowest among comparative methods.

show abstract

Section: Introductionmentioning

confidence: 99%

Occlusion Robust Wheat Ear Counting Algorithm Based on Deep Learning

2021

View full text Add to dashboard Cite

show abstract

“…The fire hazard value on the point p i is updated with the fire hazard reduction, associated to the sensor s j coverage, according to the following function, in Equation (6).…”

Section: The Optimisation Problemmentioning

confidence: 99%

“…The forest environment is composed of trees, leaves, dry, and wood, which facilitate the quick spread of fire and difficult the control in a short time [6], which also depends on the weather condition during fire ignition and the fire spread. Furthermore, the forest is an environment with several risks and uncertainty, since it involves large dimensions, irregular and remote lands, and many obstacles, such as trees, rivers, and animals.…”

Section: Introductionmentioning

confidence: 99%

Optimum Sensors Allocation for a Forest Fires Monitoring System

Azevedo

Brito

Lima

et al. 2021

Forests

View full text Add to dashboard Cite

Every year forest fires destroy millions of hectares of land worldwide. Detecting forest fire ignition in the early stages is fundamental to avoid forest fires catastrophes. In this approach, Wireless Sensor Network is explored to develop a monitoring system to send alert to authorities when a fire ignition is detected. The study of sensors allocation is essential in this type of monitoring system since its performance is directly related to the position of the sensors, which also defines the coverage region. In this paper, a mathematical model is proposed to solve the sensor allocation problem. This model considers the sensor coverage limitation, the distance, and the forest density interference in the sensor reach. A Genetic Algorithm is implemented to solve the optimisation model and minimise the forest fire hazard. The results obtained are promising since the algorithm could allocate the sensor avoiding overlaps and minimising the total fire hazard value for both regions considered.

show abstract

“…Through the application of CNN and VGGNet models, the system performs better result with 85% of accuracy and minimizes false alarm. In [13], Xu et al, propose to detect forest fire by employing deep learning methodology. Yolov5 and EfficientDet are utilized for fire detection process.…”

Section: Related Workmentioning

confidence: 99%

“…Convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory networks (LSTMs), generative adversarial Networks (GANs), radial basis function networks (RBFs), and deep belief networks (DBNs) are widely used and well-known architectures. Both traditional machine learning techniques and deep learning methodologies are employed to detect fire indoors and outdoors [11][12][13][14][15]. In this paper, we propose an efficient forest fire detection model by using traditional machine learning algorithms, deep learning models, hybrid deep learning methodologies and object detection techniques (OD) in order to demonstrate the most successful system by comparing the performance of them.…”

Section: Introductionmentioning

confidence: 99%

Comprehensive Analysis of Forest Fire Detection using Deep Learning Models and Conventional Machine Learning Algorithms

Kukuk

Kilimci

2021

International Journal of Computational and Experimental Science and Engineering

View full text Add to dashboard Cite

Forest fire detection is a very challenging problem in the field of object detection. Fire detection-based image analysis have advantages such as usage on wide open areas, the possibility for operator to visually confirm presence, intensity and the size of the hazards, lower cost for installation and further exploitation. To overcome the problem of fire detection in outdoors, deep learning and conventional machine learning based computer vision techniques are employed to determine the fire detection when indoor fire detection systems are not capable. In this work, we propose a comprehensive analysis of forest fire detection using conventional machine learning algorithms, object detection techniques, deep and hybrid deep learning models. The contribution of this work to the literature is to analyze different classification and object detection techniques in more details that is not addressed before in order to detect forest fire. Experiment results demonstrate that convolutional neural networks outperform other methods with 99.32% of accuracy result.

show abstract

A Forest Fire Detection System Based on Ensemble Learning

Cited by 369 publications

References 31 publications

Occlusion Robust Wheat Ear Counting Algorithm Based on Deep Learning

Occlusion Robust Wheat Ear Counting Algorithm Based on Deep Learning

Optimum Sensors Allocation for a Forest Fires Monitoring System

Comprehensive Analysis of Forest Fire Detection using Deep Learning Models and Conventional Machine Learning Algorithms

Contact Info

Product

Resources

About