مقاله انگلیسی رایگان در مورد شبکه تشخیص آتش سوزی سبک وزن غیر موقت – IEEE 2019

Convolutional neural networks (CNNs) have been recently applied to tackle a variety of computer vision problems. However, because of its high computational cost, careful considerations are required to design cost-effective CNNs. In this paper, we propose a CNN inspired by MobileNet for fire detection in surveillance systems. In the proposed network, color features emphasized by the channel multiplier are extracted through depthwise separable convolution, and squeeze and excitation modules further increase the representation of the channel-wise convolution. Custom Swish is used as an activation function to limit exceedingly high weights from the effects of the channel multiplier. Our proposed network achieves 95.44% accuracy for fire detection, which is higher than those achieved other existing networks. Furthermore, the number of parameters used is 38.50% fewer than that of MobileNetV2, the smallest among other networks. We believe that using the proposed CNN, CNN-based surveillance systems could be implemented in lightweight devices without using expensive dedicated processors.

Introduction

Fires can occur anywhere, at any time, and if they are not detected early, they can cause severe damages to property and people. Surveillance systems consisting of multiple CCTVs can be very useful in detecting fires because they are designed to monitor the surroundings 24 hours a day. Furthermore, they can be very useful to monitor fires in a wide range of areas, including inaccessible areas. Consequently, there has been a huge demand for intelligent video-based fire monitoring systems that can alert people to respond quickly by processing and analyzing video streams in real-time. A video-based fire detection system can inform an operator by analyzing videos from CCTVs without using heat, smoke, or flame sensors. Owing to the significant development of video analysis, video signals from CCTVs can be automatically analyzed and can provide alarms to surveillance personnel to enable quick response. Traditional vision-based fire detection methods use handcrafted features, such as color, motion, and texture. Prior studies [1]–[4] detected fire by making full use of color features because fire is generally brighter and has higher contrast than other objects. Ko et al. [1] detected specific fire regions from their color and, then employed a model using wavelet coefficients to detect fire with a support vector machine (SVM) classifier.