Pith. sign in

REVIEW 4 cited by

Benchmarking Robustness of 3D Point Cloud Recognition Against Common Corruptions

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2201.12296 v1 pith:QZ7UKJ6U submitted 2022-01-28 cs.LG cs.AIcs.CV

Jiachen Sun , Qingzhao Zhang , Bhavya Kailkhura , Zhiding Yu , Chaowei Xiao , Z. Morley Mao This is my paper
classification cs.LGcs.AIcs.CV
keywords cloudpointrobustnesscorruptionsmodelnet40-ccommoncorruptionrecognition
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Deep neural networks on 3D point cloud data have been widely used in the real world, especially in safety-critical applications. However, their robustness against corruptions is less studied. In this paper, we present ModelNet40-C, the first comprehensive benchmark on 3D point cloud corruption robustness, consisting of 15 common and realistic corruptions. Our evaluation shows a significant gap between the performances on ModelNet40 and ModelNet40-C for state-of-the-art (SOTA) models. To reduce the gap, we propose a simple but effective method by combining PointCutMix-R and TENT after evaluating a wide range of augmentation and test-time adaptation strategies. We identify a number of critical insights for future studies on corruption robustness in point cloud recognition. For instance, we unveil that Transformer-based architectures with proper training recipes achieve the strongest robustness. We hope our in-depth analysis will motivate the development of robust training strategies or architecture designs in the 3D point cloud domain. Our codebase and dataset are included in https://github.com/jiachens/ModelNet40-C

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Exploiting Vision Language Model for Training-Free 3D Point Cloud OOD Detection via Graph Score Propagation

    cs.CV 2025-06 conditional novelty 6.0 of 10

    Graph Score Propagation propagates ID prototype scores across a KNN graph of VLM text and 3D point cloud features, with prompt clustering and self-trained negative prompts, improving zero-shot and few-shot 3D OOD detection.

  2. Point-Selection Fine-Tuning Framework for Robust Point Cloud Classification

    cs.CV 2026-07 conditional novelty 5.0 of 10

    PSFT couples minimally-influential point selection with prompt tuning and stochastic feature filtering, cutting corruption error on ModelNet-C and ModelNet40-C across four pre-trained 3D backbones.

  3. OmniVec2 -- A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning

    cs.CV 2025-07 conditional novelty 5.0 of 10

    A shared-backbone transformer with pairwise modality training reports top results across 25 datasets spanning 12 modalities.

  4. Deep Learning For Point Cloud Denoising: A Survey

    cs.CV 2025-08 unverdicted novelty 4.0 of 10

    A survey of deep learning point cloud denoising, proposing a taxonomy of outlier removal and surface restoration.

Pith tools