Chinese Journal of Liquid Crystals and Displays, Volume. 37, Issue 11, 1498(2022)
Remote sensing landslide target recognition based on attention fusion
Fig. 1. Landslide identification network structure of attention fusion
Fig. 7. Comparison images before and after improved self-attentional processing
Fig. 8. Comparison images before and after convolutional block attention processing
Table 1. Dataset category
Table 2. Comparison of segmentation results
View tableView in ArticleTable 2. Comparison of segmentation results
Model Precision/% Recall/% Accuracy/% F1/% mIoU/% mPA/% U-net 96.23 94.92 94.13 95.57 81.60 86.50 DeeplabV3+ 96.50 95.37 94.81 95.94 84.35 89.11 FCN 97.36 96.61 96.16 96.70 90.81 93.66 ours 96.81 95.94 95.22 96.17 85.83 90.11
Table 3. Model complexity comparison
View tableView in ArticleTable 3. Model complexity comparison
Model Parameter/M Model size/MB GFLOPs U-net 31.0 124.3 218.6 DeeplabV3+ 54.5 219.8 82.7 FCN 134.3 537.1 190.3 ours 10.7 43.2 87.5
Table 4. Performance evaluation of different improvement strategies
View tableView in ArticleTable 4. Performance evaluation of different improvement strategies
Model SA CBAM F1/% mIoU/% mPA/% U-net 95.57 81.60 86.50 1 √ 96.00 84.78 90.36 2 √ 95.93 84.35 89.65 ours √ √ 96.17 85.83 90.11
Get Citation
Copy Citation Text
Yu WANG, Peng ZHANG, Kai-yue SUN, Xue-hong SUN, Li-ping LIU. Remote sensing landslide target recognition based on attention fusion[J]. Chinese Journal of Liquid Crystals and Displays, 2022, 37(11): 1498
Category: Research Articles
Received: Apr. 20, 2022
Accepted: --
Published Online: Nov. 3, 2022
The Author Email: Li-ping LIU (liuliping8186@126.com)
