Two-Step Active Learning for Instance Segmentation with Uncertainty and Diversity Sampling

TL;DR

This paper introduces a two-step active learning method combining uncertainty and diversity sampling to improve instance segmentation models efficiently, significantly reducing labeling costs and increasing performance on real-world datasets.

Contribution

The study proposes a novel, simple post-hoc active learning algorithm that effectively combines uncertainty and diversity sampling for instance segmentation, outperforming existing methods.

Findings

Increases labeling efficiency fivefold on overhead imagery dataset

Delivers superior performance across various datasets

Simple and easy-to-implement algorithm

Abstract

Training high-quality instance segmentation models requires an abundance of labeled images with instance masks and classifications, which is often expensive to procure. Active learning addresses this challenge by striving for optimum performance with minimal labeling cost by selecting the most informative and representative images for labeling. Despite its potential, active learning has been less explored in instance segmentation compared to other tasks like image classification, which require less labeling. In this study, we propose a post-hoc active learning algorithm that integrates uncertainty-based sampling with diversity-based sampling. Our proposed algorithm is not only simple and easy to implement, but it also delivers superior performance on various datasets. Its practical application is demonstrated on a real-world overhead imagery dataset, where it increases the labeling efficiency fivefold.