muskie82/MonoGS

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

2149 stars · 231 forks observed · 2026-08-28

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

MonoGS is a dense SLAM system that applies 3D Gaussian Splatting to monocular, stereo, and RGB-D camera tracking and mapping, presented at CVPR 2024. It runs live (around 3fps) and unifies tracking, mapping, and high-quality novel-view rendering using Gaussians as the sole 3D representation.

Use cases

• run monocular SLAM with 3D Gaussian Splatting on TUM RGB-D sequences
• reconstruct 3D scenes from an RGB-D or stereo camera in real time
• estimate camera trajectories while building splat-based dense maps
• render novel views of a live-reconstructed scene
• benchmark SLAM tracking and mapping on Replica and TUM datasets

When to choose

• you need a research-grade Gaussian Splatting SLAM baseline for monocular or RGB-D input
• you want high-fidelity rendering combined with camera tracking in one system
• you are evaluating novel view synthesis and trajectory estimation on standard SLAM datasets

When to avoid

• you need a production SLAM stack with long-term support and a permissive license
• you lack a CUDA-capable NVIDIA GPU
• you need real-time performance well above 3-10fps on modest hardware

Facets

application · maturity active

computer-vision graphics simulation robotics computer-vision graphics python slam gaussian-splatting monocular-slam rgb-d 3d-reconstruction cvpr-2024 research-code linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="muskie82/MonoGS")