Search:
Match:
1 results

Analysis

This paper addresses the challenge of fine-grained object detection in remote sensing images, specifically focusing on hierarchical label structures and imbalanced data. It proposes a novel approach using balanced hierarchical contrastive loss and a decoupled learning strategy within the DETR framework. The core contribution lies in mitigating the impact of imbalanced data and separating classification and localization tasks, leading to improved performance on fine-grained datasets. The work is significant because it tackles a practical problem in remote sensing and offers a potentially more robust and accurate detection method.
Reference

The proposed loss introduces learnable class prototypes and equilibrates gradients contributed by different classes at each hierarchical level, ensuring that each hierarchical class contributes equally to the loss computation in every mini-batch.