Spacecraft Recovery & Remote Sensing, Volume. 45, Issue 5, 79(2024)
Infrared Dim Target Detection Based on Time-Space Domain Feature Fusion
Figures & Tables(14)
![[in Chinese]](/Images/icon/loading.gif){kind=icon}
Table 1. Performance comparison of time-space domain fusion modules
View tableView in ArticleTable 1. Performance comparison of time-space domain fusion modules
模型 精确率 召回率 mAP_0.5 参数量 模型大小/MB YOLOv5s_3d 0.937 0.941 0.929 7045142 14.17 YOLOv5s 0.930 0.931 0.905 7022326 14.11
Table 2. Experiment results of attention mechanism ablation
View tableView in ArticleTable 2. Experiment results of attention mechanism ablation
模型 精确率 召回率 mAP_0.5 YOLOv5s 0.930 0.931 0.905 YOLOv5s_3d 0.937 0.941 0.929 YOLOv5s_3d_SE 0.950 0.948 0.941 YOLOv5s_3d_ECA 0.947 0.947 0.942 YOLOv5s_3d_CBAM 0.948 0.946 0.936 YOLOv5s_3d_CA （本文提出的方法） 0.952 0.952 0.940
Table 3. Comparative experiment results
View tableView in ArticleTable 3. Comparative experiment results
模型 精确率 召回率 mAP_0.5 参数量 推理速度/ms YOLOv5s 0.930 0.931 0.905 7022326 6.60 YOLOv5m 0.936 0.937 0.928 20861016 7.23 YOLOv5L 0.944 0.947 0.934 46119048 10.70 本文提出的方法 0.952 0.952 0.940 7228192 7.57
Tools
Get Citation
Copy Citation Text
Shuwei CUI, Wenbo WU. Infrared Dim Target Detection Based on Time-Space Domain Feature Fusion[J]. Spacecraft Recovery & Remote Sensing, 2024, 45(5): 79
Paper Information
Category:
Received: Jan. 30, 2024
Accepted: --
Published Online: Nov. 13, 2024
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20