SMART-PC: Skeletal Model Adaptation for Robust Test-Time Training in Point Clouds

2025-05-26

AI Summary

To address performance degradation in 3D point cloud classification under test-time distribution shifts (e.g., noise, occlusion), this paper proposes a lightweight, backpropagation-free test-time adaptation framework. Methodologically, it leverages pretrained skeletal representations to extract robust geometric features and achieves millisecond-level adaptation solely by online updating of BatchNorm statistics. It introduces the first skeleton-guided, gradient-free test-time training paradigm—eliminating parameter updates entirely and relying exclusively on statistic-driven adaptation. Evaluated on ModelNet40-C, ShapeNet-C, and ScanObjectNN-C, our method achieves state-of-the-art accuracy while attaining 128 FPS—3.2× faster inference than MATE—demonstrating an optimal trade-off between high accuracy and real-time efficiency.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Test-Time Training (TTT) has emerged as a promising solution to address distribution shifts in 3D point cloud classification. However, existing methods often rely on computationally expensive backpropagation during adaptation, limiting their applicability in real-world, time-sensitive scenarios. In this paper, we introduce SMART-PC, a skeleton-based framework that enhances resilience to corruptions by leveraging the geometric structure of 3D point clouds. During pre-training, our method predicts skeletal representations, enabling the model to extract robust and meaningful geometric features that are less sensitive to corruptions, thereby improving adaptability to test-time distribution shifts. Unlike prior approaches, SMART-PC achieves real-time adaptation by eliminating backpropagation and updating only BatchNorm statistics, resulting in a lightweight and efficient framework capable of achieving high frame-per-second rates while maintaining superior classification performance. Extensive experiments on benchmark datasets, including ModelNet40-C, ShapeNet-C, and ScanObjectNN-C, demonstrate that SMART-PC achieves state-of-the-art results, outperforming existing methods such as MATE in terms of both accuracy and computational efficiency. The implementation is available at: https://github.com/AliBahri94/SMART-PC.

Problem

• Addresses distribution shifts in 3D point cloud classification
• Reduces computational cost of test-time adaptation
• Enhances robustness to corruptions via skeletal representations

Innovation

• Skeleton-based framework for robust point cloud adaptation
• Eliminates backpropagation for real-time test-time training
• Updates BatchNorm statistics to maintain high efficiency