# pq-yang/MatAnyone

[CVPR 2025] MatAnyone: Stable Video Matting with Consistent Memory Propagation

Repository: https://github.com/pq-yang/MatAnyone
Canonical: https://ross.abutalabs.com/products/matanyone
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-03-04T15:52:40+00:00

## Health v2 (maintenance only)
Score: 54/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 70, release rhythm 40, longevity 41
- inputs: {"age_days": 577, "days_push": 182, "days_rel": 563, "gap_med": 0, "n_releases_24m": 2}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1605, forks 113 (observed 2026-08-28T04:05:10.366079+00:00)

## What it is
MatAnyone is a CVPR 2025 human video matting framework that extracts alpha mattes of target people from video using consistent memory propagation. It supports target assignment and provides inference code, a Gradio demo, Hugging Face integration, training code, and the YouTubeMatte evaluation benchmark.

## Use cases
- remove or replace video backgrounds of people
- extract alpha mattes from video footage
- segment a specific person across video frames
- green-screen-free video compositing
- benchmark video matting models on YouTubeMatte
- run video matting in a browser demo

## When to choose
- you need temporally stable human video matting with fine boundary details
- you want to mat a specific target person in multi-person scenes
- you need a research-grade model with training and evaluation code

## When to avoid
- you need real-time matting on low-end hardware without a GPU
- you need general object matting rather than humans
- you want a maintained production successor - consider MatAnyone 2

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, image-processing, video-processing, computer-vision
- domain: computer-vision, deep-learning, machine-learning
- platform: python, cross-platform
- tags: video-matting, alpha-matting, background-removal, segmentation, cvpr-2025, gradio-demo, huggingface, video, gpu

## Member repositories
- pq-yang/MatAnyone (main) score 54

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:10.366079+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 2026-08-30T03:52:17.511072+00:00, confidence not recorded.
  - readme: https://github.com/pq-yang/MatAnyone (fetched 2026-08-28T04:05:10.366079+00:00, sha 55ac1e7b4b8d)
- Data as of 2026-08-30T08:39:29.467469+00:00.