chensjtu/GaussianObject

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1183 stars · 84 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

GaussianObject is a research framework for high-quality 3D object reconstruction from as few as four input images using Gaussian splatting, including a COLMAP-free mode. It combines visual hull initialization, floater elimination, and a diffusion-based Gaussian repair model, published at SIGGRAPH Asia 2024.

Use cases

• reconstruct a 3D object from only 4 photos
• novel view synthesis of captured objects
• 3D reconstruction without COLMAP camera poses
• repair sparse-view Gaussian splatting with diffusion models
• research on few-shot 3D reconstruction
• render high-quality object views from casually captured images

When to choose

• you need 3D object reconstruction from very few images
• you want a state-of-the-art sparse-view Gaussian splatting research baseline
• you lack camera pose estimation tooling and need COLMAP-free reconstruction

When to avoid

• you need a production-ready 3D scanning pipeline with support
• you require a permissively licensed project - no license is specified
• you need full scene reconstruction rather than single objects
• you lack a GPU or ML environment

Facets

library · maturity active

machine-learning deep-learning image-processing computer-vision graphics simulation computer-vision graphics artificial-intelligence deep-learning python cross-platform gaussian-splatting 3d-reconstruction novel-view-synthesis diffusion-models research-code few-shot siggraph linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="chensjtu/GaussianObject")