Abstract:In the realm of remote sensing, targets of interest span a range of categories. However, their distribution is not always uniform. Certain categories substantially outnumber others, resulting in what's termed a "long-tailed distribution" in remote sensing imagery. This imbalanced distribution often biases a classifier's focus toward the more abundant (head) classes, at the detriment of the less-represented (tail) classes. Such biases undermine the classifier's generalization performance, particularly in the context of remote sensing image classification (RSIC). While existing mitigation approaches such as resampling, reweighting, and transfer learning offer some respite, they often miss out on in-depth knowledge refinement, rendering them less effective for severe long-tailed RSIC scenarios. To counter these challenges, we introduce DECOR, a dynamic decoupling (DD) and multiobjective optimization framework (MOOF). Within DECOR, the feature extractor and classifier are dynamically decoupled, promoting superior feature representation and classifier training. Then, a multiobjective optimization approach is proposed to delve deeper, refining feature representation at the knowledge level using learnable feature centroids (LFCs) coupled with masked world knowledge learning (WKL). Moreover, to combat the pronounced effects of sample imbalance on classifier training, we employ a class-balanced resampling technique paired with a parameter-efficient adapter, which sharpens the classifier's decision boundary and bridges the gap between representation and classification. DECOR's efficacy is validated through comprehensive experiments on several datasets, including the NWPU-RESISC45-LT (NWPU-LT), AID-LT, and our self-built BIT-AFGR50-LT. Experimental results demonstrate DECOR's marked enhancement in performance on long-tailed datasets. Our source code is available at: https://github.com/ChloeeGrace/DECOR.

Open world long-tailed data classification through active distribution optimization

Open World Classification with Adaptive Negative Samples.

Distributed Active Learning.

Deep Active Learning in the Open World

Tackling Long-Tailed Category Distribution under Domain Shifts.

Towards Realistic Long-tailed Semi-supervised Learning in an Open World

Long-Tailed Learning as Multi-Objective Optimization

Towards Distribution-Agnostic Generalized Category Discovery

Learning to Augment Distributions for Out-of-Distribution Detection

A Systematic Review on Long-Tailed Learning

Open-Sampling: Exploring Out-of-Distribution Data for Re-balancing Long-tailed Datasets

Open Long-Tailed Recognition In A Dynamic World

DECOR: Dynamic Decoupling and Multi-Objective Optimization for Long-tailed Remote Sensing Image Classification

Long-Tailed Partial Label Learning Via Dynamic Rebalancing

Learning from Neighbors: Category Extrapolation for Long-Tail Learning

Alleviating Long-Tailed Image Classification via Dynamical Classwise Splitting

Distribution-Balanced Loss for Multi-label Classification in Long-Tailed Datasets

Robust Long-Tailed Learning Via Label-Aware Bounded CVaR

Adaptive Embedding and Distribution Re-margin for Long-Tail Recognition

Label-Noise Learning with Intrinsically Long-Tailed Data

DECOR: Dynamic Decoupling and Multiobjective Optimization for Long-Tailed Remote Sensing Image Classification