Topology-Aware Modeling for Unsupervised Simulation-to-Reality Point Cloud Recognition

TL;DR

This paper introduces a topology-aware framework for unsupervised domain adaptation in 3D point cloud recognition, effectively bridging the simulation-to-reality gap by leveraging global topological features and a robust self-training strategy.

Contribution

The novel TAM framework captures global topological information and models local geometric relations through self-supervised learning, improving Sim2Real point cloud recognition.

Findings

Consistent improvements over state-of-the-art methods on three benchmarks.

Effective mitigation of the domain gap using topology-aware features.

Enhanced robustness through combined contrastive learning and self-training.

Abstract

Learning semantic representations from point sets of 3D object shapes is often challenged by significant geometric variations, primarily due to differences in data acquisition methods. Typically, training data is generated using point simulators, while testing data is collected with distinct 3D sensors, leading to a simulation-to-reality (Sim2Real) domain gap that limits the generalization ability of point classifiers. Current unsupervised domain adaptation (UDA) techniques struggle with this gap, as they often lack robust, domain-insensitive descriptors capable of capturing global topological information, resulting in overfitting to the limited semantic patterns of the source domain. To address this issue, we introduce a novel Topology-Aware Modeling (TAM) framework for Sim2Real UDA on object point clouds. Our approach mitigates the domain gap by leveraging global spatial topology, characterized by low-level, high-frequency 3D structures, and by modeling the topological relations of local geometric features through a novel self-supervised learning task. Additionally, we propose an advanced self-training strategy that combines cross-domain contrastive learning with self-training, effectively reducing the impact of noisy pseudo-labels and enhancing the robustness of the adaptation process. Experimental results on three public Sim2Real benchmarks validate the effectiveness of our TAM framework, showing consistent improvements over state-of-the-art methods across all evaluated tasks. The source code of this work will be available at https://github.com/zou-longkun/TAG.git.