# facebookresearch/vggsfm

VGGSfM: Visual Geometry Grounded Deep Structure From Motion

Repository: https://github.com/facebookresearch/vggsfm
Canonical: https://ross.abutalabs.com/products/vggsfm
Language: Python
License: NOASSERTION
License Family: other
Last push: 2025-03-11T17:11:24+00:00

## Health v2 (maintenance only)
Score: 29/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 10, release rhythm 35, longevity 61
- inputs: {"age_days": 862, "days_push": 540, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1421, forks 116 (observed 2026-08-28T04:04:40.596145+00:00)

## What it is
VGGSfM is a deep learning-based Structure from Motion pipeline from Meta AI and Oxford VGG that recovers camera poses and 3D point clouds from image collections. It is a Python/PyTorch research library with pretrained checkpoints, supporting video sequences, dynamic object masking, dense depth prediction, and Gaussian splatting export.

## Use cases
- reconstruct 3d scenes from a set of photos
- estimate camera poses from images
- generate point clouds from video frames
- recover camera trajectory from a video
- prepare camera poses for gaussian splatting
- filter moving objects in sfm reconstruction
- export dense depth maps from images

## When to choose
- you need deep-learning-based SfM that handles many frames or dynamic scenes
- you want camera pose estimation for downstream 3D tasks like Gaussian splatting
- you want a research-grade pipeline with pretrained models and a demo

## When to avoid
- you need a lightweight classical SfM tool like COLMAP without GPU/deep learning dependencies
- you need a production system with long-term support guarantees
- you cannot run CUDA/PyTorch on your machine

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, image-processing, deep-learning
- domain: computer-vision, machine-learning, deep-learning, graphics
- platform: python
- tags: structure-from-motion, camera-pose-estimation, 3d-reconstruction, point-cloud, pytorch, research, gpu, linux, macos

## Member repositories
- facebookresearch/vggsfm (main) score 29

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:40.596145+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:37:45.357111+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/vggsfm (fetched 2026-08-28T04:04:40.596145+00:00, sha ca9849347a4c)
- Data as of 2026-08-30T08:39:29.467469+00:00.