New CBEN dataset aids machine learning for cloud-affected remote sensing images

Researchers have introduced CloudyBigEarthNet (CBEN), a new multimodal dataset designed to improve machine learning models' robustness in remote sensing under cloudy conditions. Traditional methods often exclude cloudy images, limiting their applicability in time-sensitive scenarios like disaster response. CBEN pairs optical and radar satellite imagery, including occluded images, to train and evaluate models that are less affected by cloud cover. Experiments show that adapting existing methods to incorporate cloudy optical data during training significantly improves performance on cloudy test cases. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Enables more reliable remote sensing applications by improving model performance in cloudy conditions.

RANK_REASON The cluster contains an academic paper introducing a new dataset.

Read on arXiv cs.CV →

• CBEN
• CloudyBigEarthNet
• arXiv
• Marco Stricker

• paper
• other