VladimirYugay/Gaussian-SLAM

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1182 stars · 89 forks observed · 2026-08-28

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

A research implementation of a dense RGBD SLAM system that uses 3D Gaussian Splatting as its scene representation to photorealistically reconstruct and render real-world scenes. It processes posed RGBD keyframe streams and produces a real-time renderable 3D Gaussian map.

Use cases

• reconstruct a photorealistic 3D scene from an RGBD video stream
• run SLAM with Gaussian Splatting on Replica, TUM, ScanNet, or ScanNet++ datasets
• render reconstructed real-world scenes in real time
• benchmark against state-of-the-art dense SLAM methods
• research 3D Gaussian scene representations for SLAM
• reproduce paper metrics for RGBD SLAM

When to choose

• you need photo-realistic dense SLAM from RGBD input
• you want a renderable 3D Gaussian map as SLAM output
• you are doing computer vision or robotics research on SLAM and 3D reconstruction
• you have a CUDA GPU and want to evaluate on standard SLAM benchmarks

When to avoid

• you need monocular-only SLAM without depth input
• you need a production-ready, deterministic SLAM system
• you lack a modern NVIDIA GPU
• you need lightweight real-time SLAM on embedded hardware

Facets

library · maturity active

computer-vision graphics simulation machine-learning computer-vision robotics graphics python slam gaussian-splatting 3d-reconstruction rgbd photorealistic-rendering research-code research linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="VladimirYugay/Gaussian-SLAM")