Active fire detection in Landsat-8 imagery: A large-scale dataset and a deep-learning study

Pereira, Gabriel Henrique de Almeida; Fusioka, Andre Minoro; Nassu, Bogdan Tomoyuki; Minetto, Rodrigo

doi:10.1016/j.isprsjprs.2021.06.002

Cited by 95 publications

(51 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed method's architecture is a fully convolutional framework in which the output is a pixel-by-pixel map with the same size as the input image. The proposed network differs from previous AFD architectures in satellite imagery [33], which used a simple U-Net. In the proposed architecture, a combination of convolution features with different kernel sizes was implemented for improving the accuracy of AFD.…”

Section: Network Architecturementioning

confidence: 99%

“…A new large-scale dataset for AFD was recently published by De Almeida et al [33]. This dataset contained image patches of 256 × 256 pixels, depicting the wildfires in several locations around the world, and was extracted from the Landsat-8 images from August to September 2020.…”

Section: Landsat-8 Active Fire Datasetmentioning

confidence: 99%

“…We selected the intersection between the three generated ground truths. As suggested in [33], we only used the SWIR2, SWIR1, and Blue bands in this study (Figure 1).…”

Section: Landsat-8 Active Fire Datasetmentioning

confidence: 99%

“…These networks have many trainable parameters that require many labeled samples for proper estimation. To our knowledge, De Almeida et al [33] provided the first large-scale dataset for AFD. Our study's primary distinction was in developing a new efficient CNN architecture and experimenting with different network input types.…”

Section: Comparison Analysismentioning

confidence: 99%

“…Thanks to the release of the large-scale dataset for AFD by de Almeida et al [33], this paper proposed a deep encoder-decoder network, namely "MultiScale-Net", for pixel-level localization of active fire in Landsat-8 imagery. This study developed a high-performance CNN architecture for AFD to identify individual fire pixels and even single ones detached from a large fire zone.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Active Fire Detection from Landsat-8 Imagery Using Deep Multiple Kernel Learning

et al. 2022

View full text Add to dashboard Cite

Active fires are devastating natural disasters that cause socio-economical damage across the globe. The detection and mapping of these disasters require efficient tools, scientific methods, and reliable observations. Satellite images have been widely used for active fire detection (AFD) during the past years due to their nearly global coverage. However, accurate AFD and mapping in satellite imagery is still a challenging task in the remote sensing community, which mainly uses traditional methods. Deep learning (DL) methods have recently yielded outstanding results in remote sensing applications. Nevertheless, less attention has been given to them for AFD in satellite imagery. This study presented a deep convolutional neural network (CNN) “MultiScale-Net” for AFD in Landsat-8 datasets at the pixel level. The proposed network had two main characteristics: (1) several convolution kernels with multiple sizes, and (2) dilated convolution layers (DCLs) with various dilation rates. Moreover, this paper suggested an innovative Active Fire Index (AFI) for AFD. AFI was added to the network inputs consisting of the SWIR2, SWIR1, and Blue bands to improve the performance of the MultiScale-Net. In an ablation analysis, three different scenarios were designed for multi-size kernels, dilation rates, and input variables individually, resulting in 27 distinct models. The quantitative results indicated that the model with AFI-SWIR2-SWIR1-Blue as the input variables, using multiple kernels of sizes 3 × 3, 5 × 5, and 7 × 7 simultaneously, and a dilation rate of 2, achieved the highest F1-score and IoU of 91.62% and 84.54%, respectively. Stacking AFI with the three Landsat-8 bands led to fewer false negative (FN) pixels. Furthermore, our qualitative assessment revealed that these models could detect single fire pixels detached from the large fire zones by taking advantage of multi-size kernels. Overall, the MultiScale-Net met expectations in detecting fires of varying sizes and shapes over challenging test samples.

show abstract

Section: Network Architecturementioning

confidence: 99%

Section: Landsat-8 Active Fire Datasetmentioning

confidence: 99%

“…We selected the intersection between the three generated ground truths. As suggested in [33], we only used the SWIR2, SWIR1, and Blue bands in this study (Figure 1).…”

Section: Landsat-8 Active Fire Datasetmentioning

confidence: 99%

Section: Comparison Analysismentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations