In remote sensing applications, pixel-level cloud masks are indispensable in various specific tasks. Consequently, cloud detection is typically categorized as a semantic segmentation task within the domain of RS image processing. It aims to identify the presence or absence of clouds on a per-pixel basis. However, existing fully supervised cloud detection methods rely on massive pixel-wise annotations, which are expensive and time-consuming. Weakly supervised cloud detection has recently received extensive attention to alleviate the annotation burden. Here are some reference implementations on Python and Pytorch.
WDCD: Accurate cloud detection in high-resolution remote sensing imagery by weakly supervised deep learning link
Matting: Generative adversarial training for weakly supervised cloud matting link
FCD: Weakly-supervised cloud detection with fixed-point GANs link
SNMD: Weakly-supervised cloud detection and effective cloud removal for remote sensing images link