SmokeNet: Satellite Smoke Scene Detection Using Convolutional Neural Network with Spatial and Channel-Wise Attention

Abstract

A variety of environmental analysis applications have been advanced by the use of satellite remote sensing. Smoke detection based on satellite imagery is imperative for wildfire detection and monitoring. However, the commonly used smoke detection methods mainly focus on smoke discrimination from a few specific classes, which reduces their applicability in different regions of various classes. To this end, in this paper, we present a new large-scale satellite imagery smoke detection benchmark based on Moderate Resolution Imaging Spectroradiometer (MODIS) data, namely USTC_SmokeRS, consisting of 6225 satellite images from six classes (i.e., cloud, dust, haze, land, seaside, and smoke) and covering various areas/regions over the world. To build a baseline for smoke detection in satellite imagery, we evaluate several state-of-the-art deep learning-based image classification models. Moreover, we propose a new convolution neural network (CNN) model, SmokeNet, which incorporates spatial and channel-wise attention in CNN to enhance feature representation for scene classification. The experimental results of our method using different proportions (16%, 32%, 48%, and 64%) of training images reveal that our model outperforms other approaches with higher accuracy and Kappa coefficient. Specifically, the proposed SmokeNet model trained with 64% training images achieves the best accuracy of 92.75% and Kappa coefficient of 0.9130. The model trained with 16% training images can also improve the classification accuracy and Kappa coefficient by at least 4.99% and 0.06, respectively, over the state-of-the-art models.

A Review on Early Forest Fire Detection Systems Using Optical Remote Sensing

“…Deep Learning methods have also been recently applied for fire and smoke detection from multispectral satellite images. Ba et al [ 102 ] presented a new large-scale satellite imagery dataset based on MODIS data, namely USTC_SmokeRS, consisting of 6225 satellite images from six classes (i.e., cloud, dust, haze, land, seaside, and smoke) and covering various areas/regions over the world. Using this dataset, they evaluated several state-of-the-art deep learning-based image classification models for smoke detection and proposed SmokeNet , a new CNN model that incorporated spatial- and channel-wise attention in CNN to enhance feature representation for scene classification.…”

RETRACTED: Attention-Based Deep Feature Fusion for the Scene Classification of High-Resolution Remote Sensing Images

“…Generally speaking, the cross-entropy loss is frequently applied to the scene classification of HRRSI, since it can evaluate the difference between the probability distribution of true labels and that of predicted labels [51,52], which may increase the discriminative ability of the CNN. Equation 4shows the cross-entropy loss function.…”

Analysis of Multifractal and Organization/Order Structure in Suomi-NPP VIIRS Normalized Difference Vegetation Index Series of Wildfire Affected and Unaffected Sites by Using the Multifractal Detrended Fluctuation Analysis and the Fisher–Shannon Analysis

“…Deeper investigations need to be conducted at different spatial-temporal scales in conjunction with the local, regional, and global levels in order to assess the impact of fires on the vegetation communities [ 9 ]. Fortunately, the development of satellite remote sensing provides an excellent means of the continuous observation for the natural biomass [ 10 , 11 ]. Massive long-term data records are captured via multiple satellite sensors, of which the Suomi National Polar-Orbiting Partnership (Suomi-NPP) satellites were launched on October 28, 2011 as the new generation system to undertake the mission of the previous Moderate Resolution Imaging Spectroradiometer (MODIS) of Earth Observing System (EOS) satellites, which can further provide continuous data records and observations.…”