Guardado en:
Detalles Bibliográficos
Autores principales: Huang, Binbin, Duan, Haobin, Zhao, Yiqun, Zhao, Zibo, Ma, Yi, Gao, Shenghua
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2510.20776
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915635198427136
author Huang, Binbin
Duan, Haobin
Zhao, Yiqun
Zhao, Zibo
Ma, Yi
Gao, Shenghua
author_facet Huang, Binbin
Duan, Haobin
Zhao, Yiqun
Zhao, Zibo
Ma, Yi
Gao, Shenghua
contents We introduce Cupid, a generative 3D reconstruction framework that jointly models the full distribution over both canonical objects and camera poses. Our two-stage flow-based model first generates a coarse 3D structure and 2D-3D correspondences to estimate the camera pose robustly. Conditioned on this pose, a refinement stage injects pixel-aligned image features directly into the generative process, marrying the rich prior of a generative model with the geometric fidelity of reconstruction. This strategy achieves exceptional faithfulness, outperforming state-of-the-art reconstruction methods by over 3 dB PSNR and 10% in Chamfer Distance. As a unified generative model that decouples the object and camera pose, Cupid naturally extends to multi-view and scene-level reconstruction tasks without requiring post-hoc optimization or fine-tuning.
format Preprint
id arxiv_https___arxiv_org_abs_2510_20776
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CUPID: Generative 3D Reconstruction via Joint Object and Pose Modeling
Huang, Binbin
Duan, Haobin
Zhao, Yiqun
Zhao, Zibo
Ma, Yi
Gao, Shenghua
Computer Vision and Pattern Recognition
We introduce Cupid, a generative 3D reconstruction framework that jointly models the full distribution over both canonical objects and camera poses. Our two-stage flow-based model first generates a coarse 3D structure and 2D-3D correspondences to estimate the camera pose robustly. Conditioned on this pose, a refinement stage injects pixel-aligned image features directly into the generative process, marrying the rich prior of a generative model with the geometric fidelity of reconstruction. This strategy achieves exceptional faithfulness, outperforming state-of-the-art reconstruction methods by over 3 dB PSNR and 10% in Chamfer Distance. As a unified generative model that decouples the object and camera pose, Cupid naturally extends to multi-view and scene-level reconstruction tasks without requiring post-hoc optimization or fine-tuning.
title CUPID: Generative 3D Reconstruction via Joint Object and Pose Modeling
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.20776