Active Learning for Cloud Detection (ALCD) software, used to generate classification datasets with any set of categories.
Source codeOverview
The Active Learning for Cloud Detection (ALCD) software was initially designed to generate reference cloud masks which may be used to validate operational cloud masks, such as those generated by MAJA. ALCD iterative learning needs to select some points for each class and validate/correct some erroneous points after each step.
Thus, it enables to generate reference land masks for any classification. It has for example been used to generate a dataset for hydrological surfaces on Sentinel 2 images, using classes such as snow, ice, water, turbid water, etc.
Authors:
- Louis Baetens
- Camille Desjardins
- Olivier Hagolle
- Santiago Pena-Luque