Forest smoke detection based on deep learning and background modeling

Wang, Guohua; Li, Juncong; Zheng, Yongsen; Qi, Long; Gu, Wanrong

doi:10.1109/icpics50287.2020.9202287

Cited by 14 publications

(5 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [17], a dynamic background modeling mechanism was applied for improving the performance of an SSD detector using a MobileNet backbone. Considering the motion characteristic of smoke objects in video sequences, the ViBe algorithm separates the dynamic foreground objects from the stationary background in the image.…”

Section: B Object Detectionmentioning

confidence: 99%

Wildfire Detection With Deep Learning—A Case Study for the CICLOPE Project

Gonçalves,

Brandão,

Ferreira

2024

IEEE Access

View full text Add to dashboard Cite

In recent years, Portugal has seen wide variability in wildfire damage associated to high unpredictability of climatic events such as severe heatwaves and drier summers. Therefore, timely and accurate detection of forest and rural fires is of great importance for successful fire containment and suppression efforts, as wildfires exponentially increase the spread rate from the moment of ignition. In the field of early smoke detection, the CICLOPE project currently trailblazes in the employment of a network of Remote Acquisition Towers for wildfire prevention and observation, along with a rule-based automatic smoke detection system, covering over 2, 700, 000 hectares of wildland and rural area in continental Portugal. However, the inherent challenges of automatic smoke detection raise issues of high false alarm rates that affect the system's prediction quality and overwhelm the Management and Control Centers with numerous false alarms. The research work presented in this paper evaluates the potential improvement in wildfire smoke detection accuracy and specificity using deep learning-based architectures. It proposes a solution based on a Dual-Channel CNN that can be deployed as a secondary prediction confirmation layer to further refine the CICLOPE automatic smoke detection system. The proposed solution takes advantage of the high true alarm coverage of the current detection system by taking only the predicted alarm images and respective bounding box coordinates as inputs. The Dual-Channel network combines state-of-the-art DenseNet architecture with a novel detail selective network with spatial and channel attention modules trained separately with image data obtained from CICLOPE, fusing the extracted features from both networks in a concatenation layer. The results demonstrate that the proposed Dual-Channel CNN outperforms both single-channel networks achieving an accuracy of 99.7% and a low false alarm rate of 0.20% when re-examining the alarms produced by the CICLOPE surveillance system.

show abstract

Section: B Object Detectionmentioning

confidence: 99%

Wildfire Detection With Deep Learning—A Case Study for the CICLOPE Project

Gonçalves,

Brandão,

Ferreira

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The node sensor system using cameras for various functions to detect forest fires are summarized in Table 1. Others related work on the forest fire monitoring system are presented in [9]- [25]. In this work, two cameras have been employed in the system to help the user confirm the presence of smoke when the smoke trigger system is detected.…”

Section: Introductionmentioning

confidence: 99%

High accuracy sensor nodes for a peat swamp forest fire detection using ESP32 camera

Hamzah

Dalimin

Som

et al. 2022

IJ-ICT

View full text Add to dashboard Cite

<span>The </span><span lang="MS">use of smoke sensors in high-precision and low-cost forest fire detection kits needs to be developed immediately to assist the authorities in monitoring forest fires especially in remote areas more efficiently and systematically. The implementation of automatic reclosing operation allows the fire detector kit to distinguish between real smoke and non-real smoke successfully. This has profitably reduced kit errors when detecting fires and in turn prevent the users from receiving incorrect messages. However, using a smoke sensor with automatic reclosing operation has not been able to optimize the accuracy of identifying the actual smoke due to the working sensor node situation is difficult to predict and sometimes unexpected such as the source of smoke received. Thus, to further improve the accuracy when detecting the presence of smoke, the system is equipped with two digital cameras that can capture and send pictures of fire smoke to the users. The system gives the users choice of three interesting options if they want the camera to capture and send pictures to them, namely request, smoke trigger and movement for security purposes. In all cases, users can request the system to send pictures at any time. The system equipped with this camera shows the accuracy of smoke detection by confirming the actual smoke that has been detected through images sent in the user’s Telegram channel and on the Graphical User Interface (GUI) display. As a comparison of the system before and after using this camera, it was found that the system that uses the camera gives advantage to the users in monitoring fire smoke more effectively and accurately.</span>

show abstract

“…Since AlexNet 3 won the first prize in the ImageNet Competition in 2012, deep learning method has been widely applied in image and computer vision. The newest smoke detection method based on deep learning 4 used the effective convolutional neural network to extract image features automatically and achieved better detection accuracy. However, few works are proposed for high-accuracy location of smoke in the field of air pollution source monitoring, especially in the aspect of factory smoke pollution detection.…”

Section: Introductionmentioning

confidence: 99%

Real-time Factory Smoke Detection Based on Two-stage Relation-guided Algorithm

Wang

Yin

2021

Preprint

View full text Add to dashboard Cite

Recently, using image sensing devices to analyze air quality has attracted much attention of researchers. To keep real-time factory smoke under universal social supervision, this paper proposes a mobile-platform-running efficient smoke detection algorithm based on image analysis techniques. Since most smoke images in real scenes have challenging variances, it’s difficult for existing object detection methods. To this end, we introduce the two-stage smoke detection (TSSD) algorithm based on the lightweight framework, in which the prior knowledge and contextual information are modeled into the relation-guided module to reduce the smoke search space, which can therefore significantly improve the shortcomings of the single-stage method. Experimental results show that the TSSD algorithm can robustly improve the detection accuracy of the single-stage method and has good compatibility for different image resolution inputs. Compared with various state-of-the-art detection methods, the accuracy AP mean of the TSSD model reaches 59.24%, even surpassing the current detection model Faster R-CNN. In addition, the detection speed of our proposed model can reach 50 ms (20 FPS), which meets the real-time requirements, and can be deployed in the mobile terminal carrier. This model can be widely used in some scenes with smoke detection requirements, providing great potential for practical environmental applications.

show abstract

Forest smoke detection based on deep learning and background modeling

Cited by 14 publications

References 8 publications

Wildfire Detection With Deep Learning—A Case Study for the CICLOPE Project

Wildfire Detection With Deep Learning—A Case Study for the CICLOPE Project

High accuracy sensor nodes for a peat swamp forest fire detection using ESP32 camera

Real-time Factory Smoke Detection Based on Two-stage Relation-guided Algorithm

Contact Info

Product

Resources

About