CanoVerse: 3D Object Scalable Canonicalization and Dataset for Generation and Pose

Li Jin 1∗, Yuchen Yang 1∗, Weikai Chen 2†, Yujie Wang 3†, Dehao Hao 4, Tanghui Jia 5, Yingda Yin 2, Zeyu HU 2, Runze Zhang 2, Keyang Luo 2, Li Yuan 5, Long Quan 4, Xin Wang 2, Xueying Qin 1† · 2026

Read CanoVerse: 3D Object Scalable Canonicalization and Dataset for Generation and Pose by Li Jin 1∗, Yuchen Yang 1∗, Weikai Chen 2†, Yujie Wang 3†, Dehao Hao 4, Tanghui Jia 5, Yingda Yin 2, Zeyu HU 2, Runze Zhang 2, Keyang Luo 2, Li Yuan 5, Long Quan 4, Xin Wang 2, Xueying Qin 1† (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-07144.txt.

Description

Jin and colleagues build CanoVerse, a standard 3D data set of 320K objects across 1,156 categories, an order of magnitude larger than prior work. Raw 3D assets arrive with random whole spin, which blocks pose-steady making and stable way meaning. Their lining up system cuts lining up from minutes to seconds per object through compact guess making and light human checks. At this scale, way meaning becomes learnable, lifting making safety, cross-modal finding, and zero-shot aim guessing.

Catalog record

CREATORLi Jin 1∗, Yuchen Yang 1∗, Weikai Chen 2†, Yujie Wang 3†, Dehao Hao 4, Tanghui Jia 5, Yingda Yin 2, Zeyu HU 2, Runze Zhang 2, Keyang Luo 2, Li Yuan 5, Long Quan 4, Xin Wang 2, Xueying Qin 1†
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37019
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.07144

Browse the full library · About the Tower of Babel