Exploring 3D Dataset Pruning

TL;DR

This paper introduces a novel method for pruning 3D datasets that balances class coverage and accuracy metrics, improving efficiency and performance in 3D data training scenarios.

Contribution

It formulates 3D dataset pruning as an expected risk approximation problem and proposes a representation-aware, bias-invariant pruning approach with adjustable trade-offs.

Findings

Improves both overall and mean accuracy across multiple 3D datasets.

Effectively handles long-tail class distributions in 3D data.

Provides a flexible pruning method adaptable to different downstream needs.

Abstract

Dataset pruning has been widely studied for 2D images to remove redundancy and accelerate training, while particular pruning methods for 3D data remain largely unexplored. In this work, we study dataset pruning for 3D data, where its observed common long-tail class distribution nature make optimization under conventional evaluation metrics Overall Accuracy (OA) and Mean Accuracy (mAcc) inherently conflicting, and further make pruning particularly challenging. To address this, we formulate pruning as approximating the full-data expected risk with a weighted subset, which reveals two key errors: coverage error from insufficient representativeness and prior-mismatch bias from inconsistency between subset-induced class weights and target metrics. We propose representation-aware subset selection with per-class retention quotas for long-tail coverage, and prior-invariant teacher supervision using calibrated soft labels and embedding-geometry distillation. The retention quota also serves as a switch to control the OA-mAcc trade-off. Extensive experiments on 3D datasets show that our method can improve both metrics across multiple settings while adapting to different downstream preferences. Our code is available at https://github.com/XiaohanZhao123/3D-Dataset-Pruning.