Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Oct 8;22(19):7624.
doi: 10.3390/s22197624.

A Residual-Inception U-Net (RIU-Net) Approach and Comparisons with U-Shaped CNN and Transformer Models for Building Segmentation from High-Resolution Satellite Images

Batuhan Sariturk  1 , Dursun Zafer Seker  1
Affiliations

A Residual-Inception U-Net (RIU-Net) Approach and Comparisons with U-Shaped CNN and Transformer Models for Building Segmentation from High-Resolution Satellite Images

Batuhan Sariturk et al. Sensors (Basel). .

Abstract

Building segmentation is crucial for applications extending from map production to urban planning. Nowadays, it is still a challenge due to CNNs' inability to model global context and Transformers' high memory need. In this study, 10 CNN and Transformer models were generated, and comparisons were realized. Alongside our proposed Residual-Inception U-Net (RIU-Net), U-Net, Residual U-Net, and Attention Residual U-Net, four CNN architectures (Inception, Inception-ResNet, Xception, and MobileNet) were implemented as encoders to U-Net-based models. Lastly, two Transformer-based approaches (Trans U-Net and Swin U-Net) were also used. Massachusetts Buildings Dataset and Inria Aerial Image Labeling Dataset were used for training and evaluation. On Inria dataset, RIU-Net achieved the highest IoU score, F1 score, and test accuracy, with 0.6736, 0.7868, and 92.23%, respectively. On Massachusetts Small dataset, Attention Residual U-Net achieved the highest IoU and F1 scores, with 0.6218 and 0.7606, and Trans U-Net reached the highest test accuracy, with 94.26%. On Massachusetts Large dataset, Residual U-Net accomplished the highest IoU and F1 scores, with 0.6165 and 0.7565, and Attention Residual U-Net attained the highest test accuracy, with 93.81%. The results showed that RIU-Net was significantly successful on Inria dataset. On Massachusetts datasets, Residual U-Net, Attention Residual U-Net, and Trans U-Net provided successful results.

Keywords: CNN; Inception; Transformer; building segmentation; residual connections; satellite images.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Sample 256 × 256 pixel image and mask: (a) Inria dataset, (b) Massachusetts dataset.
Figure 2
Figure 2
(a) Residual connection design from the ResNet. (b) Residual connection design implemented in the study.
Figure 3
Figure 3
Attention mechanism implemented in the study [26].
Figure 4
Figure 4
Overall architecture of the RIU-Net.
Figure 5
Figure 5
The flow diagram of the modules used in the encoder path of the RIU-Net: (a) Module A, (b) Module B, (c) Module C, (d) Reduction A, and (e) Reduction B.
Figure 6
Figure 6
The flow diagram of the modules used in the bottleneck and decoder paths of the RIU-Net: (a) Module D, and (b) Upsampling module.
Figure 7
Figure 7
Evaluation metric results on Inria test set.
Figure 8
Figure 8
Evaluation metric results on Massachusetts Small test set.
Figure 9
Figure 9
Evaluation metric results on Massachusetts Large test set.
Figure 10
Figure 10
Inria test set image no. 1036 segmentation results.
Figure 11
Figure 11
Massachusetts Small test set image no. 192 segmentation results.
Figure 12
Figure 12
Massachusetts Large test set image no. 294 segmentation results.
See this image and copyright information in PMC

References

    1. Chen J., Jiang Y., Luo L., Gu Y., Wu K. Building footprint generation by integrating U-Net with deepened space module; Proceedings of the 2021 IEEE International Conference on Image Processing (ICIP); Anchorage, AK, USA. 19–22 September 2021; pp. 3847–3851.
    1. Zhang Y., Gong W., Sun J., Li W. Web-Net: A novel nest networks with ultra-hierarchical sampling for building extraction from aerial imageries. Remote Sens. 2019;11:1897. doi: 10.3390/rs11161897. - DOI
    1. Yu M., Chen X., Zhang W., Liu Y. AGs-Unet: Building Extraction Model for High Resolution Remote Sensing Images Based on Attention Gates U Network. Sensors. 2022;22:2932. doi: 10.3390/s22082932. - DOI - PMC - PubMed
    1. Wang H., Miao F. Building extraction from remote sensing images using deep residual U-Net. Eur. J. Remote Sens. 2022;55:71–85. doi: 10.1080/22797254.2021.2018944. - DOI
    1. Sun X., Zhao W., Maretto R.V., Persello C. Building outline extraction from aerial imagery and digital surface model with a frame field learning framework. Int. Arch. Photogramm. Remote Sens. Spat. Inf. 2021;43:487–493. doi: 10.5194/isprs-archives-XLIII-B2-2021-487-2021. - DOI

MeSH terms

LinkOut - more resources