TY - JOUR
T1 - Multiscale context aggregation network for building change detection using high resolution remote sensing images
AU - Dong, Jie
AU - Zhao, Wufan
AU - Wang, Shuai
N1 - Publisher Copyright:
IEEE
PY - 2021/10/18
Y1 - 2021/10/18
N2 - The existing methods of building change detection (CD) using remote sensing (RS) images are still deficient in handling scale variation and class imbalance problems, indicating a decrease in the robustness of small-object detection and pseudo-change information. Thus, a novel building CD framework called the multiscale context aggregation network (MSCANet) is proposed. The high-resolution network is integrated into the feature extracting stage to maintain high-resolution representations throughout the whole process. Then, multiscale context information is aggregated using a scale-aware feature pyramid module (FPM). Recognition performance can be improved from discriminant feature representation learning by using a channel–spatial attention module. Furthermore, a class-balanced loss is proposed to reduce the impact of class imbalance in long-tail datasets. Experimental results from using the LEVIR-CD and SZTAKI AirChange benchmark datasets prove the superiority of the MSCANet over the other baseline methods, with improved maximum F1 scores of 5.28 and 8.47, respectively.
AB - The existing methods of building change detection (CD) using remote sensing (RS) images are still deficient in handling scale variation and class imbalance problems, indicating a decrease in the robustness of small-object detection and pseudo-change information. Thus, a novel building CD framework called the multiscale context aggregation network (MSCANet) is proposed. The high-resolution network is integrated into the feature extracting stage to maintain high-resolution representations throughout the whole process. Then, multiscale context information is aggregated using a scale-aware feature pyramid module (FPM). Recognition performance can be improved from discriminant feature representation learning by using a channel–spatial attention module. Furthermore, a class-balanced loss is proposed to reduce the impact of class imbalance in long-tail datasets. Experimental results from using the LEVIR-CD and SZTAKI AirChange benchmark datasets prove the superiority of the MSCANet over the other baseline methods, with improved maximum F1 scores of 5.28 and 8.47, respectively.
KW - ITC-ISI-JOURNAL-ARTICLE
KW - n/a OA procedure
U2 - 10.1109/LGRS.2021.3121094
DO - 10.1109/LGRS.2021.3121094
M3 - Article
SN - 1545-598X
VL - 19
SP - 1
EP - 5
JO - IEEE geoscience and remote sensing letters
JF - IEEE geoscience and remote sensing letters
ER -