Generative Active Learning for Long-tailed Instance Segmentation

TL;DR

This paper introduces BSGAL, an active learning algorithm that effectively selects and utilizes generated data to improve long-tailed instance segmentation, outperforming baseline methods.

Contribution

The paper proposes BSGAL, a novel active learning method that estimates generated data contribution using gradient cache for better long-tailed segmentation performance.

Findings

BSGAL outperforms baseline approaches in experiments.

It effectively handles unlimited generated data.

Improves performance on long-tailed segmentation tasks.

Abstract

Recently, large-scale language-image generative models have gained widespread attention and many works have utilized generated data from these models to further enhance the performance of perception tasks. However, not all generated data can positively impact downstream models, and these methods do not thoroughly explore how to better select and utilize generated data. On the other hand, there is still a lack of research oriented towards active learning on generated data. In this paper, we explore how to perform active learning specifically for generated data in the long-tailed instance segmentation task. Subsequently, we propose BSGAL, a new algorithm that online estimates the contribution of the generated data based on gradient cache. BSGAL can handle unlimited generated data and complex downstream segmentation tasks effectively. Experiments show that BSGAL outperforms the baseline approach and effectually improves the performance of long-tailed segmentation. Our code can be found at https://github.com/aim-uofa/DiverGen.