Comparing the robustness of U-Net, LinkNet, and FPN towards label noise for refugee dwelling extraction from satellite imagery.
Yunya GaoStefan LangDirk TiedeGetachew Workineh GellaLorenz WendtPublished in: GHTC (2022)
Keyphrases
- satellite imagery
- road extraction
- label noise
- change detection
- remote sensing
- satellite images
- active learning
- urban areas
- land cover
- aerial images
- image analysis
- multispectral
- remote sensing images
- evolutionary algorithm
- feature selection
- noise tolerant
- noise model
- computer vision
- high quality
- training data
- travel time
- learning algorithm
- high resolution
- data streams