Optics and Precision Engineering, Volume. 31, Issue 11, 1700(2023)
Fast extraction of buildings from remote sensing images by fusion of CNN and Transformer
Figures & Tables(12)
Fig. 1. Network structure of fast extraction algorithm for buildings from remote sensing image by fusiou of CNN and Transformer
Fig. 7. Characteristic image of Example 2 and spatial probability distribution
Fig. 8. Changes of characteristic map before and after using spatial attention
Table 1. Performance comparison before and after model improvement
View tableView in ArticleTable 1. Performance comparison before and after model improvement
Algorithm OA/% IoU/% RT/ms STTNet 97.10 78.88 40.01 改进STTNet 97.11 78.98 32.67
Table 2. Comparison of various methods on the INRIA dataset
View tableView in ArticleTable 2. Comparison of various methods on the INRIA dataset
Algorithm OA/% IoU/% RT/ms U-Nets［ 16 ］96.05 74.55 108.84 Building-A-Nets［ 17 ］96.71 78.73 90.56 GMEDN［ 18 ］96.43 76.69 38.72 STTNet［ 13 ］97.10 78.88 40.01 改进STTNet 97.11 78.98 32.67
Table 3. Comparison of ablation experimental results
View tableView in ArticleTable 3. Comparison of ablation experimental results
Model OA/% IoU/% RT/ms STTNet 97.10 78.88 40.01 STTNet+（1）+（2） 97.18 78.99 43.12 STTNet+（1）+（2）+（3） 97.11 78.98 32.67
Tools
Get Citation
Copy Citation Text
Yunzuo ZHANG, Wei GUO, Cunyu WU. Fast extraction of buildings from remote sensing images by fusion of CNN and Transformer[J]. Optics and Precision Engineering, 2023, 31(11): 1700
Paper Information
Category: Information Sciences
Received: Sep. 15, 2022
Accepted: --
Published Online: Jul. 4, 2023
The Author Email: ZHANG Yunzuo (zhangyunzuo888@sina.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20