All Points Matter: Entropy-Regularized Distribution Alignment for Weakly-supervised 3D Segmentation

TL;DR

This paper introduces ERDA, a novel entropy-regularized distribution alignment method that enhances weakly supervised 3D segmentation by effectively utilizing all unlabeled data, significantly improving performance over existing approaches.

Contribution

The paper proposes a simple yet effective ERDA learning strategy combining entropy regularization and distribution alignment, outperforming prior methods in weakly supervised 3D segmentation.

Findings

ERDA achieves state-of-the-art results on multiple datasets.

ERDA outperforms fully-supervised baselines with only 1% labeled data.

The method effectively exploits all unlabeled data points for improved learning.

Abstract

Pseudo-labels are widely employed in weakly supervised 3D segmentation tasks where only sparse ground-truth labels are available for learning. Existing methods often rely on empirical label selection strategies, such as confidence thresholding, to generate beneficial pseudo-labels for model training. This approach may, however, hinder the comprehensive exploitation of unlabeled data points. We hypothesize that this selective usage arises from the noise in pseudo-labels generated on unlabeled data. The noise in pseudo-labels may result in significant discrepancies between pseudo-labels and model predictions, thus confusing and affecting the model training greatly. To address this issue, we propose a novel learning strategy to regularize the generated pseudo-labels and effectively narrow the gaps between pseudo-labels and model predictions. More specifically, our method introduces an Entropy Regularization loss and a Distribution Alignment loss for weakly supervised learning in 3D segmentation tasks, resulting in an ERDA learning strategy. Interestingly, by using KL distance to formulate the distribution alignment loss, it reduces to a deceptively simple cross-entropy-based loss which optimizes both the pseudo-label generation network and the 3D segmentation network simultaneously. Despite the simplicity, our method promisingly improves the performance. We validate the effectiveness through extensive experiments on various baselines and large-scale datasets. Results show that ERDA effectively enables the effective usage of all unlabeled data points for learning and achieves state-of-the-art performance under different settings. Remarkably, our method can outperform fully-supervised baselines using only 1% of true annotations. Code and model will be made publicly available at https://github.com/LiyaoTang/ERDA.