.png){kind=icon}
.png){kind=icon}
Return
Gradient-guided dynamic multi-scale network for camouflaged object detection
.png){kind=icon}
PRE
AI
DOI：10.1016/j.imavis.2026.106096.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A collaborative optimization framework (GDMNet) with gradient guidance and dynamic multi-scale decoding is proposed for camouflaged object detection. • A gradient-aware feature fusion module (GFFM) leverages learnable Laplacian gradient priors to enhance edge responses • A dynamic multi-scale decoder (DMD) adaptively adjusts receptive fields via a learnable scale selection mechanism. • State-of-the-art performance is achieved on three benchmark datasets, significantly outperforming 12 existing methods.
Journal
IF:
4.2
Papers:
4.0K
Citations:
6.7K