Dual Deep Learning Model for Image Based Smoke Detection

Pundir, Arun Singh; Raman, Balasubramanian

doi:10.1007/s10694-019-00872-2

Cited by 49 publications

(23 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Furthermore, knowing that the deep CNN used by Pundir et al [28] remains the best alternative for smoke detection, the performances of the proposed method are compared to [28] in this section. Our goal is not to beat the robustness criteria values achieved by this method, but to propose an alternative for smoke detection with a simpler architecture.…”

Section: Comparison Of Smoke Detection Results Using Support Vector Machine and Deep Cnnmentioning

confidence: 99%

“…The training process of the deep CNN method is summarized in Figure 12 [28]. Unlike this method, in our proposed technique of smoke detection, we only insert the chosen feature vector for classification using DBN to directly obtain the smoke and no-smoke probabilities (Figure 5).…”

Section: Comparison Of Smoke Detection Results Using Support Vector Machine and Deep Cnnmentioning

confidence: 99%

“…To the best of our knowledge, the only research paper using DBN and dealing with smoke detection is Pundir's one [28]. The difference between the proposed approach and the method developed in [27] is that we are interested only by color, motion, and energy features, and we inserted these features on the DBN classifier contrarily to the aforementioned method [28] based on motion, color, and texture analysis. For this reason, we propose, in this paper, to show the importance of using energy as a relevant smoke feature in the classification framework by comparing the performances of both methods.…”

Section: Dataset Presentation and Proposed Methodology Descriptionmentioning

confidence: 99%

See 2 more Smart Citations

An Efficient Smoke Detection Algorithm Based on Deep Belief Network Classifier Using Energy and Intensity Features

et al. 2020

View full text Add to dashboard Cite

Smoke detection plays an important role in forest safety warning systems and fire prevention. Complicated changes in the shape, texture, and color of smoke remain a substantial challenge to identify smoke in a given image. In this paper, a new algorithm using the deep belief network (DBN) is designed for smoke detection. Unlike popular deep convolutional networks (e.g., Alex-Net, VGG-Net, Res-Net, Dense-Net, and the denoising convolution neural network (DNCNN), specifically devoted to detecting smoke), our proposed end-to-end network is mainly based on DBN. Indeed, most traditional smoke detection algorithms follow the pattern recognition process which consists basically feature extraction and classification. After extracting the candidate regions, the main idea is to perform both smoke recognition and smoke-no-smoke region classification using static and dynamic smoke characteristics. However, manual smoke detection cannot meet the requirements of a high smoke detection rate and has a long processing time. The convolutional neural network (CNN)-based smoke detection methods are significantly slower due to the maxpooling operation. In addition, the training phase can take a lot of time if the computer is not equipped with a powerful graphics processing unit (GPU). Thus, the contribution of this work is the development of a preprocessing step including a new combination of features—smoke color, smoke motion, and energy—to extract the regions of interest which are inserted within a simple architecture with the deep belief network (DBN). Our proposed method is able to classify and localize reliably the smoke regions providing an interesting computation time and improved performance metrics. First, the Gaussian mixture model (GMM) is employed to capture the frames containing a large amount of motion. After applying RGB rules to smoke pixels and analyzing the energy attitude of smoke regions, extracted features are then used to feed a DBN for classification. Experimental results conducted on the publicly available smoke detection database confirm that the DBN has reached a high detection rate that exceeded an average of 96% when tested on different videos containing smoke-like objects, which make smoke recognition more challenging. The proposed methodology provided high detection ratios and low false alarms, and guaranteed robustness verified by evaluations of accuracy, F1-score, and recall for noisy and non-noisy images with and without noise.

show abstract

Section: Comparison Of Smoke Detection Results Using Support Vector Machine and Deep Cnnmentioning

confidence: 99%

Section: Comparison Of Smoke Detection Results Using Support Vector Machine and Deep Cnnmentioning

confidence: 99%

Section: Dataset Presentation and Proposed Methodology Descriptionmentioning

confidence: 99%

See 1 more Smart Citation

An Efficient Smoke Detection Algorithm Based on Deep Belief Network Classifier Using Energy and Intensity Features

et al. 2020

View full text Add to dashboard Cite

show abstract

“…At the same time, in order to avoid the local continuous nature of the aircraft landing gear image from being changed, it is necessary to filter out the cracks in the image noise. The hyperspectral sensor is used to obtain the hyperspectral image of the aircraft landing gear, and the acquired hyperspectral image is denoised, which provides relevant information for the intelligent detection of internal cracks in the aircraft landing gear image [14].…”

Section: Image Denoising Of Aircraft Landing Gearmentioning

confidence: 99%

Intelligent Detection Method for Internal Cracks in Aircraft Landing Gear Images under Multimedia Processing

Luo

Zhang

2021

Symmetry

View full text Add to dashboard Cite

In view of the lack of image enhancement processing in the traditional methods in image preprocessing, which leads to a long detection time for internal cracks in the image and poor visual effects, an intelligent detection method for internal cracks in aircraft landing gear images under multimedia processing is proposed. A spatial index structure is established based on the multimedia database, and the aircraft landing gear images in the structure are enhanced and denoised. Image segmentation is performed according to the preprocessing results, the crack foreground and the road surface background in the image are separated, and the threshold value of each image is calculated. The threshold segmentation result is used to distinguish which pixels are the areas where the cracks may exist and which pixels belong to the image background, and the judgment result realizes crack detection. The experimental results show that the crack detection time of the proposed method is shorter, the visual effect of the detection results is better, and the average of the expert score reaches 93.6 points, which verifies the effectiveness of the proposed method from both the subjective and objective aspects.

show abstract

“…Compared to geometric features (e.g., texture and shape) in images, color has greater stability and has been widely applied in computer-vision detection in the region of interest (ROI) [42][43][44] . Generally, if the ROI has its own unique color features in the image, it is certainly feasible to directly identify the ROI based on color difference 45 .…”

Section: Identification Of Riparian Zones Based On the Optimized Rgb-mentioning

confidence: 99%

Mapping soil pollution risk from riparian land-cover changes in the headwater area of an inter-basin water diversion project

Wang

Zhang

Yang

et al. 2021

Preprint

View full text Add to dashboard Cite

A novel scheme was proposed for mapping precise land-cover change in riparian zones and the resulted soil pollution risk of non-point source (NPS) from an inter-basin water diversion project on its headwater area. The scheme held three key points: 1) digital visualization of the riparian land-cover’s cyclical features was done using remote sensing data fusion of NDWI at high water level and NDVI and MNDWI at lower water level; 2) riparian zones were identified through color clustering on optimized RGB-fusion image that was picked out by applying Modified Vector-Angular Distance (MVAD) formula to measure difference between riparian and non-riparian zones in RGB space; 3) combining Export Coefficient Model (ECM) and the derived data on riparian land-cover change around Danjiangkou Reservoir in China, time-series loads of the soil NPS pollutants were estimated before and after the Middle Route of South-to-North Water Diversion Project (SNWDP) started running in 2014. The results indicated that risk of the potential soil pollution from land-cover change in the new riparian zones enlarged. Growth rates (GR) of total nitrogen (TN) and total phosphorus (TP) not only had significant spatial variation, but also were higher during initial running period of the SNWDP’s Middle Route.

show abstract

Dual Deep Learning Model for Image Based Smoke Detection

Cited by 49 publications

References 39 publications

An Efficient Smoke Detection Algorithm Based on Deep Belief Network Classifier Using Energy and Intensity Features

An Efficient Smoke Detection Algorithm Based on Deep Belief Network Classifier Using Energy and Intensity Features

Intelligent Detection Method for Internal Cracks in Aircraft Landing Gear Images under Multimedia Processing

Mapping soil pollution risk from riparian land-cover changes in the headwater area of an inter-basin water diversion project

Contact Info

Product

Resources

About