Joint multi-modal Self-Supervised pre-training in Remote Sensing: Application to Methane Source Classification

In the task of multi-modal classification, we introduce a contrastive learning framework which is based on the regular contrastive learning from computer vision.