PartLLM

A Unified Multimodal Foundation for 3D Part Segmentation

SIGGRAPH Asia 2026 (ACM Transactions on Graphics)

Zhe ZhuYiheng ZhangPeng LiZixing ZhaoHonghua ChenYaqing ZhangLe WanZhiyang Dou

Cheng Lin‡Yuan Liu†Mingqiang Wei†Wenping Wang

‡ Project lead† Corresponding authors

PaperCodeModel

Abstract

Part segmentation is a fundamental problem in computer graphics and 3D vision. Recent works have expanded 3D part segmentation beyond fixed taxonomies, but existing approaches typically only address a specific setting, such as text-guided part segmentation or point-based interaction. In this work, we argue that these settings can be unified as an intent-conditioned generative problem, where different prompts specify the desired part decomposition. To this end, we introduce PartLLM, a unified multimodal model that formulates 3D part segmentation as autoregressive semantic decomposition. Conditioned on an input shape and a user prompt, PartLLM autoregressively generates semantic part hypotheses as queries for mask prediction and feeds them to a decomposition-aware decoder that jointly predicts coherent part masks. This unified design supports text-guided part segmentation, interactive segmentation, and full-shape semantic decomposition with controllable granularity within a single model. Extensive experiments across these task settings show that PartLLM consistently outperforms task-specific baselines, demonstrating the effectiveness of unifying 3D part segmentation under an intent-conditioned generative formulation.

Method Overview

Results

Full-shape Segmentation

Granularity
Loading model…
Drag to rotate · Scroll to zoom

Text-guided Segmentation

Drag to rotate · Scroll to zoom

Interactive Segmentation

Interaction
Drag to rotate · Scroll to zoom

BibTeX

@misc{zhu2026partllmunifiedmultimodalfoundation,
    title={PartLLM: A Unified Multimodal Foundation for 3D Part Segmentation},
    author={Zhe Zhu and Yiheng Zhang and Peng Li and Zixing Zhao and Honghua Chen and Yaqing Zhang and Le Wan and Zhiyang Dou and Cheng Lin and Yuan Liu and Mingqiang Wei and Wenping Wang},
    year={2026},
    eprint={2609.25832},
    archivePrefix={arXiv},
    primaryClass={cs.CV},
    url={https://arxiv.org/abs/2609.25832}
}