MARC: Multi-Label Adaptive Retrieval Contrastive Loss for Remote Sensing Images

TL;DR

This paper introduces MACL, a contrastive learning extension for remote sensing image retrieval that addresses semantic imbalance and improves performance across multiple datasets.

Contribution

It proposes Multi-Label Adaptive Contrastive Learning with label-aware sampling, frequency-sensitive weighting, and dynamic-temperature scaling for better multi-label remote sensing retrieval.

Findings

MACL outperforms contrastive-loss baselines on benchmark datasets.

It effectively mitigates semantic imbalance in remote sensing data.

The method improves retrieval reliability in large-scale archives.

Abstract

Semantic overlap among land-cover categories, highly imbalanced label distributions, and complex inter-class co-occurrence patterns constitute significant challenges for multi-label remote-sensing image retrieval. In this article, Multi-Label Adaptive Contrastive Learning (MACL) is introduced as an extension of contrastive learning to address them. It integrates label-aware sampling, frequency-sensitive weighting, and dynamic-temperature scaling to achieve balanced representation learning across both common and rare categories. Extensive experiments on three benchmark datasets (DLRSD, ML-AID, and WHDLD), show that MACL consistently outperforms contrastive-loss based baselines, effectively mitigating semantic imbalance and delivering more reliable retrieval performance in large-scale remote-sensing archives. Code, pretrained models, and evaluation scripts will be released at https://github.com/Amna-128/MARC upon acceptance.