DM-OSVP++: One-Shot View Planning Using 3D Diffusion Models for Active RGB-Based Object Reconstruction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pan, Sicong, Jin, Liren, Huang, Xuying, Stachniss, Cyrill, Popović, Marija, Bennewitz, Maren
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916692980924416
author Pan, Sicong
Jin, Liren
Huang, Xuying
Stachniss, Cyrill
Popović, Marija
Bennewitz, Maren
author_facet Pan, Sicong
Jin, Liren
Huang, Xuying
Stachniss, Cyrill
Popović, Marija
Bennewitz, Maren
contents Active object reconstruction is crucial for many robotic applications. A key aspect in these scenarios is generating object-specific view configurations to obtain informative measurements for reconstruction. One-shot view planning enables efficient data collection by predicting all views at once, eliminating the need for time-consuming online replanning. Our primary insight is to leverage the generative power of 3D diffusion models as valuable prior information. By conditioning on initial multi-view images, we exploit the priors from the 3D diffusion model to generate an approximate object model, serving as the foundation for our view planning. Our novel approach integrates the geometric and textural distributions of the object model into the view planning process, generating views that focus on the complex parts of the object to be reconstructed. We validate the proposed active object reconstruction system through both simulation and real-world experiments, demonstrating the effectiveness of using 3D diffusion priors for one-shot view planning.
format Preprint
id arxiv_https___arxiv_org_abs_2504_11674
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DM-OSVP++: One-Shot View Planning Using 3D Diffusion Models for Active RGB-Based Object Reconstruction
Pan, Sicong
Jin, Liren
Huang, Xuying
Stachniss, Cyrill
Popović, Marija
Bennewitz, Maren
Robotics
Computer Vision and Pattern Recognition
Active object reconstruction is crucial for many robotic applications. A key aspect in these scenarios is generating object-specific view configurations to obtain informative measurements for reconstruction. One-shot view planning enables efficient data collection by predicting all views at once, eliminating the need for time-consuming online replanning. Our primary insight is to leverage the generative power of 3D diffusion models as valuable prior information. By conditioning on initial multi-view images, we exploit the priors from the 3D diffusion model to generate an approximate object model, serving as the foundation for our view planning. Our novel approach integrates the geometric and textural distributions of the object model into the view planning process, generating views that focus on the complex parts of the object to be reconstructed. We validate the proposed active object reconstruction system through both simulation and real-world experiments, demonstrating the effectiveness of using 3D diffusion priors for one-shot view planning.
title DM-OSVP++: One-Shot View Planning Using 3D Diffusion Models for Active RGB-Based Object Reconstruction
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.11674