GitHub - chaibosong/SAFPN: This paper proposes a ship detection and segmentation method based on an improved Mask R-CNN model. By constructing a bottom-up feature enhancement path in MASK R-CNN, the transfer path between the strong localization features in the shallow layer and the top-level network is shortened to improve small target detection and segmentation. Attention modules are introduced in FPN and bottom-up structures, respectively. By combining channel attention, spatial attention and self-attention, the long-distance dependencies between image pixels are captured while suppressing interfering features such as noise in complex backgrounds. Add skip connections between the same layers to fuse more features. The scalability is achieved by stacking this feature pyramid structure, which further improves the feature extraction capability of the network.

chaibosong / SAFPN Public

Notifications You must be signed in to change notification settings
Fork 2
Star 4

This paper proposes a ship detection and segmentation method based on an improved Mask R-CNN model. By constructing a bottom-up feature enhancement path in MASK R-CNN, the transfer path between the strong localization features in the shallow layer and the top-level network is shortened to improve small target detection and segmentation. Attentio…

4 stars 2 forks Branches Tags Activity

Star

Notifications

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
configs		configs
demo		demo
dist		dist
logs		logs
maskrcnn_benchmark.egg-info		maskrcnn_benchmark.egg-info
maskrcnn_benchmark		maskrcnn_benchmark
others		others
tools		tools
.flake8		.flake8
INSTALL.md		INSTALL.md
MODEL_ZOO.md		MODEL_ZOO.md
compute.py		compute.py
setup.py		setup.py