Disc3D: Automatic Curation of High-Quality 3D Dialog Data via Discriminative Object Referring

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wei, Siyuan, Wang, Chunjie, Liu, Xiao, Yan, Xiaosheng, Zhou, Zhishan, Huang, Rui
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918296431886336
author Wei, Siyuan
Wang, Chunjie
Liu, Xiao
Yan, Xiaosheng
Zhou, Zhishan
Huang, Rui
author_facet Wei, Siyuan
Wang, Chunjie
Liu, Xiao
Yan, Xiaosheng
Zhou, Zhishan
Huang, Rui
contents 3D Multi-modal Large Language Models (MLLMs) still lag behind their 2D peers, largely because large-scale, high-quality 3D scene-dialogue datasets remain scarce. Prior efforts hinge on expensive human annotation and leave two key ambiguities unresolved: viewpoint ambiguity, where spatial language presumes unknown camera poses, and object referring ambiguity, where non-exclusive descriptions blur the line between targets and distractors. We therefore present a fully automated pipeline that converts raw 3D scans into unambiguous, high-quality dialogue data at a fraction of the previous cost. By synergizing rule-based constraints with 2D MLLMs and LLMs, the pipeline enables controllable, scalable generation without human intervention. The pipeline comprises four stages: (1) meta-annotation collection harvesting object-, frame-, and scene-level captions, (2) scene graph construction with relation correction to capture proximal object relations, (3) discriminative object referring that generates exclusive and compact descriptions, and (4) multi-task data generation synthesizing diverse dialogues. Our pipeline systematically mitigates inherent flaws in source datasets and produces the final Disc3D dataset, over 2 million samples in 25K hybrid 3D scenes, spanning scene, view, and object captioning, visual grounding, and five object-centric QA tasks. Extensive experiments demonstrate that training with Disc3D yields consistent, significant improvements on both public benchmarks and our multifaceted Disc3D-QA tasks. Code, data, and models will be publicly available.
format Preprint
id arxiv_https___arxiv_org_abs_2511_18817
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Disc3D: Automatic Curation of High-Quality 3D Dialog Data via Discriminative Object Referring
Wei, Siyuan
Wang, Chunjie
Liu, Xiao
Yan, Xiaosheng
Zhou, Zhishan
Huang, Rui
Computer Vision and Pattern Recognition
3D Multi-modal Large Language Models (MLLMs) still lag behind their 2D peers, largely because large-scale, high-quality 3D scene-dialogue datasets remain scarce. Prior efforts hinge on expensive human annotation and leave two key ambiguities unresolved: viewpoint ambiguity, where spatial language presumes unknown camera poses, and object referring ambiguity, where non-exclusive descriptions blur the line between targets and distractors. We therefore present a fully automated pipeline that converts raw 3D scans into unambiguous, high-quality dialogue data at a fraction of the previous cost. By synergizing rule-based constraints with 2D MLLMs and LLMs, the pipeline enables controllable, scalable generation without human intervention. The pipeline comprises four stages: (1) meta-annotation collection harvesting object-, frame-, and scene-level captions, (2) scene graph construction with relation correction to capture proximal object relations, (3) discriminative object referring that generates exclusive and compact descriptions, and (4) multi-task data generation synthesizing diverse dialogues. Our pipeline systematically mitigates inherent flaws in source datasets and produces the final Disc3D dataset, over 2 million samples in 25K hybrid 3D scenes, spanning scene, view, and object captioning, visual grounding, and five object-centric QA tasks. Extensive experiments demonstrate that training with Disc3D yields consistent, significant improvements on both public benchmarks and our multifaceted Disc3D-QA tasks. Code, data, and models will be publicly available.
title Disc3D: Automatic Curation of High-Quality 3D Dialog Data via Discriminative Object Referring
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.18817