Data-Centric Benchmark for Label Noise Estimation and Ranking in Remote Sensing Image Segmentation

TL;DR

This paper presents a new benchmark and dataset for identifying and ranking noisy labels in remote sensing image segmentation, improving model robustness by focusing on label noise detection techniques.

Contribution

It introduces a novel data-centric benchmark, dataset, and methods leveraging uncertainty and consistency for label noise estimation in remote sensing segmentation.

Findings

Proposed methods outperform existing baselines.

Techniques effectively identify noisy labels.

Benchmark facilitates future research in label noise handling.

Abstract

High-quality pixel-level annotations are essential for the semantic segmentation of remote sensing imagery. However, such labels are expensive to obtain and often affected by noise due to the labor-intensive and time-consuming nature of pixel-wise annotation, which makes it challenging for human annotators to label every pixel accurately. Annotation errors can significantly degrade the performance and robustness of modern segmentation models, motivating the need for reliable mechanisms to identify and quantify noisy training samples. This paper introduces a novel Data-Centric benchmark, together with a novel, publicly available dataset and two techniques for identifying, quantifying, and ranking training samples according to their level of label noise in remote sensing semantic segmentation. Such proposed methods leverage complementary strategies based on model uncertainty, prediction consistency, and representation analysis, and consistently outperform established baselines across a range of experimental settings. The outcomes of this work are publicly available at https://github.com/keillernogueira/label_noise_segmentation.