Foreground-Background Contrastive Learning for Few-Shot Remote Sensing Image Scene Classification

Few-shot learning (FSL) aims to train a model with limited samples for identifying novel category samples. As for remote sensing (RS) images, complex backgrounds may lead to large intraclass differences, and the number of labeled samples is quite smaller than that of large datasets, which both influence the classification performance. To solve these issues, a foreground-background contrastive learning (FBCL) is proposed for few-shot RS image scene classification. Specifically, a foreground-background separation (FBS) module is proposed to separate features between objects and background with supervised contrastive learning (SCL), which aims to improve the ability to distinguish foreground and background regions of RS images. Moreover, a channel weight allocator is proposed to balance features of different dimensions, which can take full advantage of RS image information. Experiments on three RS datasets prove that the proposed few-shot method is able to produce superior classification results than other related approaches.

Automated summary

Few-shot learning is addressed for remote sensing image scene classification by proposing a foreground-background contrastive learning method. It includes a foreground-background separation module for distinguishing object and background features and a channel weight allocator for balancing feature dimensions. Experimental results on three remote sensing datasets demonstrate its superior classification performance compared to other approaches.