Overall framework of DEDM-Net. The network consists of 2 stages. (IMAGE)
Caption
Overall framework of DEDM-Net. The network consists of 2 stages. In the first stage, a dual-path processing scheme is employed to precisely model both the predicted target mask and the target-free background, thereby suppressing background clutter and enhancing target features. In the second stage, based on the enhanced features from the first stage, a dense nested Mamba structure is applied to achieve multilevel extraction of small target features and effective integration of global information, which improves detection performance while reducing false alarm probability.
Credit
Journal of Remote Sensing
Usage Restrictions
Credit must be given to the creator.
License
CC BY