Saved in:
Bibliographic Details
Main Authors: Song, Yizheng, Zhuang, Yiyu, Xu, Qipeng, Wang, Haixiang, Zhu, Jiahe, Tian, Jing, Zhu, Siyu, Zhu, Hao
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.17779
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917357661716480
author Song, Yizheng
Zhuang, Yiyu
Xu, Qipeng
Wang, Haixiang
Zhu, Jiahe
Tian, Jing
Zhu, Siyu
Zhu, Hao
author_facet Song, Yizheng
Zhuang, Yiyu
Xu, Qipeng
Wang, Haixiang
Zhu, Jiahe
Tian, Jing
Zhu, Siyu
Zhu, Hao
contents Single-view 3D human reconstruction has garnered significant attention in recent years. Despite numerous advancements, prior research has concentrated on reconstructing 3D models from clear, close-up images of individual subjects, often yielding subpar results in the more prevalent multi-person scenarios. Reconstructing 3D human crowd models is a highly intricate task, laden with challenges such as: 1) extensive occlusions, 2) low clarity, and 3) numerous and various appearances. To address this task, we propose CrowdGaussian, a unified framework that directly reconstructs multi-person 3D Gaussian Splatting (3DGS) representations from single-image inputs. To handle occlusions, we devise a self-supervised adaptation pipeline that enables the pretrained large human model to reconstruct complete 3D humans with plausible geometry and appearance from heavily occluded inputs. Furthermore, we introduce Self-Calibrated Learning (SCL). This training strategy enables single-step diffusion models to adaptively refine coarse renderings to optimal quality by blending identity-preserving samples with clean/corrupted image pairs. The outputs can be distilled back to enhance the quality of multi-person 3DGS representations. Extensive experiments demonstrate that CrowdGaussian generates photorealistic, geometrically coherent reconstructions of multi-person scenes.
format Preprint
id arxiv_https___arxiv_org_abs_2603_17779
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CrowdGaussian: Reconstructing High-Fidelity 3D Gaussians for Human Crowd from a Single Image
Song, Yizheng
Zhuang, Yiyu
Xu, Qipeng
Wang, Haixiang
Zhu, Jiahe
Tian, Jing
Zhu, Siyu
Zhu, Hao
Computer Vision and Pattern Recognition
Single-view 3D human reconstruction has garnered significant attention in recent years. Despite numerous advancements, prior research has concentrated on reconstructing 3D models from clear, close-up images of individual subjects, often yielding subpar results in the more prevalent multi-person scenarios. Reconstructing 3D human crowd models is a highly intricate task, laden with challenges such as: 1) extensive occlusions, 2) low clarity, and 3) numerous and various appearances. To address this task, we propose CrowdGaussian, a unified framework that directly reconstructs multi-person 3D Gaussian Splatting (3DGS) representations from single-image inputs. To handle occlusions, we devise a self-supervised adaptation pipeline that enables the pretrained large human model to reconstruct complete 3D humans with plausible geometry and appearance from heavily occluded inputs. Furthermore, we introduce Self-Calibrated Learning (SCL). This training strategy enables single-step diffusion models to adaptively refine coarse renderings to optimal quality by blending identity-preserving samples with clean/corrupted image pairs. The outputs can be distilled back to enhance the quality of multi-person 3DGS representations. Extensive experiments demonstrate that CrowdGaussian generates photorealistic, geometrically coherent reconstructions of multi-person scenes.
title CrowdGaussian: Reconstructing High-Fidelity 3D Gaussians for Human Crowd from a Single Image
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.17779