Deformation-based In-Context Learning for Point Cloud Understanding

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lin, Chengxing, Deng, Jinhong, Lei, Yinjie, Li, Wen
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917382740508672
author Lin, Chengxing
Deng, Jinhong
Lei, Yinjie
Li, Wen
author_facet Lin, Chengxing
Deng, Jinhong
Lei, Yinjie
Li, Wen
contents Recent advances in point cloud In-Context Learning (ICL) have demonstrated strong multitask capabilities. Existing approaches typically adopt a Masked Point Modeling (MPM)-based paradigm for point cloud ICL. However, MPM-based methods directly predict the target point cloud from masked tokens without leveraging geometric priors, requiring the model to infer spatial structure and geometric details solely from token-level correlations via transformers. Additionally, these methods suffer from a training-inference objective mismatch, as the model learns to predict the target point cloud using target-side information that is unavailable at inference time. To address these challenges, we propose DeformPIC, a deformation-based framework for point cloud ICL. Unlike existing approaches that rely on masked reconstruction, DeformPIC learns to deform the query point cloud under task-specific guidance from prompts, enabling explicit geometric reasoning and consistent objectives. Extensive experiments demonstrate that DeformPIC consistently outperforms previous state-of-the-art methods, achieving reductions of 1.6, 1.8, and 4.7 points in average Chamfer Distance on reconstruction, denoising, and registration tasks, respectively. Furthermore, we introduce a new out-of-domain benchmark to evaluate generalization across unseen data distributions, where DeformPIC achieves state-of-the-art performance.
format Preprint
id arxiv_https___arxiv_org_abs_2604_02845
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Deformation-based In-Context Learning for Point Cloud Understanding
Lin, Chengxing
Deng, Jinhong
Lei, Yinjie
Li, Wen
Computer Vision and Pattern Recognition
Recent advances in point cloud In-Context Learning (ICL) have demonstrated strong multitask capabilities. Existing approaches typically adopt a Masked Point Modeling (MPM)-based paradigm for point cloud ICL. However, MPM-based methods directly predict the target point cloud from masked tokens without leveraging geometric priors, requiring the model to infer spatial structure and geometric details solely from token-level correlations via transformers. Additionally, these methods suffer from a training-inference objective mismatch, as the model learns to predict the target point cloud using target-side information that is unavailable at inference time. To address these challenges, we propose DeformPIC, a deformation-based framework for point cloud ICL. Unlike existing approaches that rely on masked reconstruction, DeformPIC learns to deform the query point cloud under task-specific guidance from prompts, enabling explicit geometric reasoning and consistent objectives. Extensive experiments demonstrate that DeformPIC consistently outperforms previous state-of-the-art methods, achieving reductions of 1.6, 1.8, and 4.7 points in average Chamfer Distance on reconstruction, denoising, and registration tasks, respectively. Furthermore, we introduce a new out-of-domain benchmark to evaluate generalization across unseen data distributions, where DeformPIC achieves state-of-the-art performance.
title Deformation-based In-Context Learning for Point Cloud Understanding
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.02845