readme.md

A hierarchical ML benchmark for Sentinels, with three task levels covering all major Sentinel missions.

• level 1: preprocessing
  • S2 cloud segmentation
  • S3 cloud segmentation
• level 2: base application
  • single label LULC classification
    • S1 EuroSAT-SAR
    • S2 EuroSAT-MS
  • multi label LULC classification
    • S1 BigEarthNet-v2-SAR
    • S2 BigEarthNet-v2-MS
    • S3 LC100
  • dense LULC segmentation
    • S1 DFC2020-SAR
    • S2 DFC2020-MS
    • S3 LC100
• level 3: specialized application
  • S1 flood segmentation
  • S2 local climate zone classification
  • S3 biomass regression
  • S5P air quality regression
    • NO2
    • O3