# MCG-NJU/VideoMAE

[NeurIPS 2022 Spotlight] VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training

Repository: https://github.com/MCG-NJU/VideoMAE
Canonical: https://ross.abutalabs.com/products/videomae
Homepage: https://arxiv.org/abs/2203.12602
Language: Python
License: NOASSERTION
License Family: other
Topics: self-supervised-learning, action-recognition, video-understanding, masked-autoencoder, transformer, vision-transformer, video-transformer, mae, pytorch, video-representation-learning, video-analysis, neurips-2022
Last push: 2023-12-08T13:44:48+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1624, "days_push": 999, "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 1784, forks 171 (observed 2026-08-28T04:05:35.976085+00:00)

## What it is
Official PyTorch implementation of VideoMAE, a masked autoencoder method for data-efficient self-supervised video pre-training with video tube masking. It provides pre-trained ViT-based video models and training/finetuning code for action recognition benchmarks like Kinetics-400, Something-Something V2, UCF101, and HMDB51.

## Use cases
- pre-train video transformers self-supervised on small video datasets
- fine-tune video models for action recognition
- classify human actions in videos
- extract video representations for downstream tasks
- reproduce NeurIPS 2022 VideoMAE results
- train video models without extra labeled data

## When to choose
- you need data-efficient self-supervised video pre-training on limited data
- you want strong action recognition backbones with released checkpoints
- you are researching masked autoencoders for video

## When to avoid
- you need a permissively licensed model for commercial use (CC BY-NC 4.0)
- you need real-time video inference in production
- you work outside PyTorch

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, video-processing, computer-vision
- domain: machine-learning, computer-vision, artificial-intelligence
- platform: python
- tags: masked-autoencoder, self-supervised-learning, video-transformer, action-recognition, pre-training, pytorch, vision-transformer, video, gpu

## Member repositories
- MCG-NJU/VideoMAE (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:35.976085+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 2026-08-30T03:24:14.291764+00:00, confidence not recorded.
  - readme: https://github.com/MCG-NJU/VideoMAE (fetched 2026-08-28T04:05:35.976085+00:00, sha 303f1494ab48)
  - homepage: https://arxiv.org/abs/2203.12602 (fetched 2026-08-29T11:02:47.012475+00:00, sha 287998c80b5e)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T11:02:47.022621+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T11:02:47.026273+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T11:02:47.028096+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T11:02:47.024652+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.