Core-Set Selection for Data-efficient Land Cover Segmentation

2025-05-02

AI Summary

To address data redundancy, label noise, and high computational cost in land-cover segmentation of remote sensing imagery, this paper proposes six purely data-driven core-set selection methods, categorized into three paradigms: image-feature-only, label-distribution-only, and hybrid feature–distribution approaches. These methods leverage image clustering, label distribution optimization, and multimodal consistency modeling—requiring no auxiliary model training or human intervention. Evaluated on DFC2022, Vaihingen, and Potsdam benchmarks, models (e.g., U-Net) trained on selected subsets (30–50% of full data) achieve higher mIoU than those trained on the complete dataset, substantially outperforming random sampling. This work establishes the first systematic core-set selection framework tailored to remote sensing segmentation, empirically validating the “quality-over-quantity” data paradigm and demonstrating improved training efficiency, generalization, and robustness.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Dimensionality Reduction/Feature Selection
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The increasing accessibility of remotely sensed data and the potential of such data to inform large-scale decision-making has driven the development of deep learning models for many Earth Observation tasks. Traditionally, such models must be trained on large datasets. However, the common assumption that broadly larger datasets lead to better outcomes tends to overlook the complexities of the data distribution, the potential for introducing biases and noise, and the computational resources required for processing and storing vast datasets. Therefore, effective solutions should consider both the quantity and quality of data. In this paper, we propose six novel core-set selection methods for selecting important subsets of samples from remote sensing image segmentation datasets that rely on imagery only, labels only, and a combination of each. We benchmark these approaches against a random-selection baseline on three commonly used land cover classification datasets: DFC2022, Vaihingen, and Potsdam. In each of the datasets, we demonstrate that training on a subset of samples outperforms the random baseline, and some approaches outperform training on all available data. This result shows the importance and potential of data-centric learning for the remote sensing domain. The code is available at https://github.com/keillernogueira/data-centric-rs-classification/.

Problem

• Selecting important subsets for land cover segmentation
• Improving data efficiency in remote sensing tasks
• Reducing dataset size without losing performance

Innovation

• Core-set selection methods for data efficiency
• Image and label-based sample subset selection
• Outperforms random and full dataset training