# google-deepmind/kinetics-i3d

Convolutional neural network model for video classification trained on the Kinetics dataset.

Repository: https://github.com/google-deepmind/kinetics-i3d
Canonical: https://ross.abutalabs.com/products/kinetics-i3d
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2019-09-12T14:04:07+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": 3318, "days_push": 2547, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1838, forks 467 (observed 2026-08-28T04:05:42.823268+00:00)

## What it is
A repository of pre-trained Inflated 3D Convnet (I3D) models for video action classification, trained on the Kinetics dataset, released alongside the CVPR 2017 'Quo Vadis' paper. It includes TensorFlow/Sonnet sample code for building the two-stream model and running inference with pre-trained RGB and flow checkpoints.

## Use cases
- classify human actions in videos
- get a pretrained video model for transfer learning
- fine-tune I3D on UCF101 or HMDB51
- run video action recognition inference with TensorFlow
- reproduce results from the Quo Vadis action recognition paper

## When to choose
- you need a strong pretrained video action recognition backbone for research or fine-tuning
- you work in TensorFlow/Sonnet and want the original I3D checkpoints
- you want to reproduce or build on the CVPR 2017 Kinetics paper

## When to avoid
- you need PyTorch or modern framework support
- you want actively maintained code with recent updates
- you need general video processing rather than action classification
- you require production-grade video classification pipelines

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, computer-vision, video-processing
- domain: machine-learning, computer-vision
- platform: python
- tags: i3d, video-classification, action-recognition, kinetics-dataset, pretrained-models, tensorflow, sonnet, research, video

## Member repositories
- google-deepmind/kinetics-i3d (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:42.823268+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 2026-08-30T03:18:16.387196+00:00, confidence not recorded.
  - readme: https://github.com/google-deepmind/kinetics-i3d (fetched 2026-08-28T04:05:42.823268+00:00, sha 1752d3578127)
- Data as of 2026-08-30T08:39:29.467469+00:00.