CrowdGaussian: Reconstructing High-Fidelity 3D Gaussians for Human Crowd from a Single Image

Yizheng Song; Yiyu Zhuang; Qipeng Xu; Haixiang Wang; Jiahe Zhu; Jing Tian; Siyu Zhu; Hao Zhu

cs.CVMar 18, 2026

CrowdGaussian: Reconstructing High-Fidelity 3D Gaussians for Human Crowd from a Single Image

Authors: Yizheng Song, Yiyu Zhuang, Qipeng Xu, Haixiang Wang, Jiahe Zhu, Jing Tian, Siyu Zhu, Hao Zhu

Organizations: Nanjing University, China · State Key Laboratory of Novel Software Technology, China · Fudan University, China

Abstract

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.

Figures & tables

Figure 2: Overview of the proposed CrowdGaussian framework. Our pipeline operates in two stages. In Stage 1 , we first estimate SMPL-X parameters and segment individuals from the input image. These occluded crops are processed by our LORM to hallucinate complete geometries, assembling an initial coarse multi-person 3DGS scene. In Stage 2 , we render this coarse scene into RGB images and normal maps. Our CrowdRefiner leverages these cues to generate high-fidelity pseudo-ground truths, which are then distilled back into the 3D Gaussians via differentiable rendering, significantly enhancing local details and overall sharpness.

Figure 3: The Large Occluded Human Reconstruction Model (LORM). (a) Architecture. LORM takes an occluded image and a template to reconstruct a complete 3D human. To preserve priors while enabling efficient adaptation, we freeze pre-trained backbones and inject trainable LoRA exclusively into the transformer. (b) Self-Supervised Training. We employ a Teacher-Student framework where the teacher generates clean pseudo-GTs from complete images. These signals guide the student to hallucinate complete geometries from occluded inputs via self-distillation, achieving robustness without external 3D supervision.

Figure 4: Architecture of CrowdRefiner, our single-step diffusion refiner for Crowd 3D Gaussians enhancement. Given a coarse rendering and its corresponding SMPL normal map as geometric prior, CrowdRefiner generates a high-fidelity refined output. The model is fine-tuned from SD-Turbo with a LoRA-finetuned VAE decoder and a trainable PoseNet, while the VAE encoder remains frozen. Zoom-in regions highlight significant improvements in local details such as hair and clothing textures.

Figure 5: Effect of Self-Calibrated Learning (SCL). Without SCL (middle), the model tends to over-refine, causing facial distortions and artifacts. With SCL (right), structural integrity is preserved while details are enhanced, demonstrating adaptive refinement enabled by mixed identity supervision during training.

Figure 6: Single-shot multi-person 3D reconstruction. While the LHM baseline ( middle ) struggles with artifacts and missing details in occluded regions, our method ( right ) generates complete geometry and high-fidelity textures. Zoom-ins highlight our superior recovery of faces and hands, validating robustness to occlusion and low-resolution inputs.

Figure 7: Qualitative comparison on occluded in-the-wild images. Mesh-based methods (PSHuman, SyncHuman) fail to recover complete geometry. 3DGS-based approaches (IDOL, LHM) produce transparency artifacts or incoherent and distorted textures in missing regions. In contrast, our method reconstructs complete 3D geometry and texture, demonstrating robustness to real-world occlusion.

Method	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
IDOL [ 58 ]	18.063	0.919	0.994
LHM [ 29 ]	18.171	0.918	1.012
LORM	18.566	0.923	0.956
LORM+CrowdRefiner	18.619	0.931	0.914

Table 1: Quantitative comparison on occluded human reconstruction (THuman2.1) with randomly applied masks.

Method	Occlusion Ratio	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
	20%	18.196	0.921	0.978
IDOL	40%	17.412	0.914	1.027
	60%	16.667	0.909	1.063
	20%	17.945	0.919	1.006
LHM	40%	17.709	0.916	1.026
	60%	17.551	0.915	1.037

Table 2: Quantitative comparison under increasing occlusion ratios (20%–60%) on THuman2.1.

Figure 8: Robustness to resolution degradation. The leftmost column shows the original image, followed by inputs downsampled by $2\times$ and $4\times$ . Subsequent columns compare reconstructions from IDOL, LHM, and Ours. Even from significantly downsampled inputs, our approach consistently recovers high-fidelity geometry and sharp textures, demonstrating superior robustness to input resolution degradation compared to baselines.

SCL	Normal	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
$-$	$-$	20.013	0.888	0.141
$-$	✓	20.130	0.892	0.138
✓	$-$	20.382	0.896	0.129
✓	✓	20.790	0.901	0.122

Table 3: Ablation study on the SCL strategy and geometric conditioning inputs for the refiner. Results demonstrate that both SCL and the additional normal map input contribute to consistent improvements in generation quality, providing higher-fidelity supervision for the subsequent stage.

Figure 9: Visualization of the multi-stage occlusion mask generation process. From left to right: original image, with keypoint-based elliptical occlusions applied, further corrupted by irregular Bézier-based erasures, and finally smoothed via morphological operations.

Figure 10: Ablation on geometric conditioning using SMPL normal maps. Left: Coarse input. Middle: Result without normal map conditioning — suffers from hand collapse and facial distortion. Right: Result with normal map — preserves correct hand structure and facial geometry. Zoom-ins show significant improvement in local detail fidelity and structural coherence.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 11: Qualitative comparison on occluded human reconstruction (THuman2.1). Input images are occluded via Bézier curves or rectangular masks.

Figure 12: More reconstruction results on diverse in-the-wild multi-person images.

Figure 13: Comparison with the 2D inpainting + LHM pipeline.

3D Human Reconstruction

3D Human Reconstruction recovers human shape, pose, appearance, and motion from images, video, or other sensor data.

See all 103 papers

Contrastive Multi-Modal Hypergraph Reasoning for 3D Crowd Mesh Recovery
Apr 1, 2026Minghao Sun, Chongyang Xu, Yitao Xie +23D Reconstruction 3D Human Reconstruction
Tianjin University · Nanyang Technological University · Sichuan University
Human Interaction-Aware 3D Reconstruction from a Single Image
Apr 7, 2026Gwanghyun Kim, Junghun James Kim, Suh Yoon Jeon +23D Human Reconstruction Single-View 3D Reconstruction
Dept. of Electrical and Computer Engineering, Seoul National University, Republic of Korea · IPAI, Seoul National University, Republic of Korea · INMC, Seoul National University, Republic of Korea
Feed-Forward Multi-view Multi-person Reconstruction with Contrastive Human-Aware 3D Representation
Sep 1, 2026Yuanwang Yang, Buzhen Huang, Zongxuan Ren +2Multi-View 3D Reconstruction 3D Human Reconstruction
College of Intelligence and Computing, Tianjin University, 135 Yaguan Road, Tianjin, 300350, Tianjin, China.

Apr 1, 2026J/K move · Enter open · S save

Contrastive Multi-Modal Hypergraph Reasoning for 3D Crowd Mesh Recovery

Minghao Sun, Chongyang Xu, Yitao Xie, Buzhen Huang+1

Tianjin University · Nanyang Technological University · Sichuan University

3D Reconstruction 3D Human Reconstruction Hypergraph Neural Networks Multimodal Learning

Open paper AskPDF arXiv

Multi-person 3D reconstruction is pivotal for real-world interaction analysis, yet remains challenging due to severe occlusions and depth ambiguity. Current approaches typically rely on single-modality inputs, which inherently lack geometric guidance. Furthermore, these methods often reconstruct subjects in isolation, neglecting the collective group context essential for resolving ambiguities in crowded scenes. To address these limitations, we propose Contrastive Multi-modal Hypergraph Reasoning to synergize semantic, geometric, and pose cues for crowd reconstruction. We first initialize robust node representations by combining RGB features, geometric priors, and occlusion-aware incomplete poses. Additionally, we introduce a pelvis depth indicator as a global spatial anchor, aligning visual features with a metric-scale-agnostic depth ordering. Subsequently, we construct a shared-topology hypergraph that moves beyond pairwise constraints to model higher-order crowd dynamics. To improve feature fusion, we design a hypergraph-based contrastive learning scheme that jointly enhances intra-modal discriminability and enforces cross-modal orthogonality. This mechanism enables the network to propagate global context effectively, allowing it to infer missing information even under severe occlusion. Extensive experiments on the Panoptic and GigaCrowd benchmarks confirm that our method achieves new state-of-the-art performance. Code and pre-trained models are available at https://github.com/SunMH-try/CoMHR.

Abstract

3D Human Reconstruction

Explore similar work

Contrastive Multi-Modal Hypergraph Reasoning for 3D Crowd Mesh Recovery

Human Interaction-Aware 3D Reconstruction from a Single Image

Feed-Forward Multi-view Multi-person Reconstruction with Contrastive Human-Aware 3D Representation