Lookalike3D: Seeing Double in 3D

Chandan Yeshwanth Angela Dai · 2026

Read Lookalike3D: Seeing Double in 3D by Chandan Yeshwanth Angela Dai (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-2603-24713.txt.

Description

Yeshwanth and Dai introduce the task of lookalike object detection in indoor 3D scenes. Repeated objects are a rich but overlooked cue in real scenes. The task classifies object pairs as identical, similar, or different, from multiview images. Their Lookalike3D model is a multiview image transformer that draws on strong priors from large image foundation models. To support the task, they built 3DTwins, a dataset of 76,000 hand-annotated pairs from ScanNet++. The method improves IoU by 104 percent over baselines and helps downstream reconstruction and part co-segmentation.

Catalog record

CREATORChandan Yeshwanth Angela Dai
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE44830
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.24713

Browse the full library · About the Tower of Babel