Fusing frozen DINOv2 features into DeepLabV3 improves flood segmentation mIoU by 6.36 points on FloodNet, while the U-Net variant gains only 0.79 points over vanilla U-Net.
The limitations of traditional surveying techniques in a forested environment
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Leveraging Self-Supervised Features for Efficient Flooded Region Identification in UAV Aerial Images
Fusing frozen DINOv2 features into DeepLabV3 improves flood segmentation mIoU by 6.36 points on FloodNet, while the U-Net variant gains only 0.79 points over vanilla U-Net.