Effects of SAR Resolution in Automatic Building Segmentation Using CNN

Wangiyana, Sandhi; Samczyński, Piotr; Gromek, Artur

doi:10.1109/spsympo51155.2020.9593636

Cited by 1 publication

(2 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For aerial imagery, DeepLab v3 [29], PSPNet [30], and Feature Pyramid Network (FPN) [31] are commonly used. Based on a previous study [32], we used the FPN architecture combined with the EfficientNet B4 backbone. EfficientNet is a family of CNN models generated using compound scaling to determine an optimal network size [33].…”

Section: Segmentation Modelmentioning

confidence: 99%

“…This is effective in homogenous areas, but in applications requiring high-frequency information such as edges, filters that can adapt to local texture can better preserve information in heterogeneous areas [39]. A previous study [32] has shown a slight performance gain by applying Low Pass filters with varying strength on the UNet model. In this research, we inspected the use of well-known adaptive speckle filters as a form of data augmentation, namely Enhanced Lee Filter (eLee), Frost Filter, and Gamma Maximum A Posteriori (GMAP) Filter.…”

Section: Speckle Filtersmentioning

confidence: 99%

See 1 more Smart Citation

Data Augmentation for Building Footprint Segmentation in SAR Images: An Empirical Study

2022

View full text Add to dashboard Cite

Building footprints provide essential information for mapping, disaster management, and other large-scale studies. Synthetic Aperture Radar (SAR) provides consistent data availability over optical images owing to its unique properties, which consequently makes it more challenging to interpret. Previous studies have demonstrated the success of automated methods using Convolutional Neural Networks to detect buildings in Very High Resolution (VHR) SAR images. However, the scarcity of such datasets that are available to the public can limit research progress in this field. We explored the impact of several data augmentation (DA) methods on the performance of building detection on a limited dataset of SAR images. Our results show that geometric transformations are more effective than pixel transformations. The former improves the detection of objects with different scale and rotation variations. The latter creates textural changes that help differentiate edges better, but amplifies non-object patterns, leading to increased false positive predictions. We experimented with applying DA at different stages and concluded that applying similar DA methods in training and inference showed the best performance compared with DA applied only during training. Some DA can alter key features of a building’s representation in radar images. Among them are vertical flips and quarter circle rotations, which yielded the worst performance. DA methods should be used in moderation to prevent unwanted transformations outside the possible object variations. Error analysis, either through statistical methods or manual inspection, is recommended to understand the bias presented in the dataset, which is useful in selecting suitable DAs. The findings from this study can provide potential guidelines for future research in selecting DA methods for segmentation tasks in radar imagery.

show abstract