# facebookresearch/sapiens

High-resolution models for human tasks.

Repository: https://github.com/facebookresearch/sapiens
Canonical: https://ross.abutalabs.com/products/sapiens
Homepage: https://about.meta.com/realitylabs/codecavatars/sapiens/
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-05-26T19:50:55+00:00

## Health v2 (maintenance only)
Score: 61/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 84, release rhythm 35, longevity 53
- inputs: {"age_days": 750, "days_push": 99, "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 5418, forks 321 (observed 2026-08-28T04:09:17.871313+00:00)

## What it is
Sapiens is a family of foundation models from Meta Reality Labs for human-centric vision tasks including 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. The models are pretrained on 300 million in-the-wild human images, natively support 1K high-resolution inference, and scale from 0.3 to 2 billion parameters.

## Use cases
- estimate 2D human pose from images
- segment human body parts in photos
- predict monocular depth for people in images
- predict surface normals of human bodies
- fine-tune human vision models on custom datasets
- extract high-resolution human features for avatar creation

## When to choose
- you need state-of-the-art human-centric vision models with strong generalization
- you work with high-resolution (1K) human imagery
- you have limited labeled data and want strong pretrained backbones
- you need multiple human vision tasks from one model family

## When to avoid
- you need general object detection or non-human vision tasks
- you lack GPU resources for billion-parameter models
- you need a lightweight model for edge or mobile deployment
- you require a permissive license without restrictions

## Facets
- artifact type: library
- maturity: active
- function: computer-vision, image-processing, machine-learning, deep-learning
- domain: computer-vision, deep-learning, artificial-intelligence
- platform: python
- tags: pose-estimation, depth-estimation, surface-normals, segmentation, human-vision, eccv-2024, pretrained-models, gpu, linux

## Member repositories
- facebookresearch/sapiens (main) score 61

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:17.871313+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 2026-08-29T17:56:57.183047+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/sapiens (fetched 2026-08-28T04:09:17.871313+00:00, sha 1ae4aedd1225)
  - homepage: https://about.meta.com/realitylabs/codecavatars/sapiens/ (fetched 2026-08-29T08:52:17.431866+00:00, sha c225bed202f3)
- Data as of 2026-08-30T08:39:29.467469+00:00.