MEAL: Manifold Embedding-based Active Learning

TL;DR

This paper introduces MEAL, a novel active learning method for image segmentation in autonomous driving, combining manifold embedding and entropy-based uncertainty to select informative image patches, leading to improved annotation efficiency.

Contribution

The paper presents a new pool-based active learning approach that uses manifold embedding and entropy to select image patches, enhancing segmentation performance with less labeled data.

Findings

Outperforms state-of-the-art active learning methods on CamVid and Cityscapes datasets.

Effective in reducing annotation effort for image segmentation tasks.

Demonstrates improved segmentation accuracy with fewer labeled samples.

Abstract

Image segmentation is a common and challenging task in autonomous driving. Availability of sufficient pixel-level annotations for the training data is a hurdle. Active learning helps learning from small amounts of data by suggesting the most promising samples for labeling. In this work, we propose a new pool-based method for active learning, which proposes promising patches extracted from full image, in each acquisition step. The problem is framed in an exploration-exploitation framework by combining an embedding based on Uniform Manifold Approximation to model representativeness with entropy as uncertainty measure to model informativeness. We applied our proposed method to the autonomous driving datasets CamVid and Cityscapes and performed a quantitative comparison with state-of-the-art baselines. We find that our active learning method achieves better performance compared to previous methods.