google-deepmind/kinetics-i3d
Inflated 3D Convnet (I3D) CNN for video/action classification trained on the Kinetics dataset, released by DeepMind.
Not currently ranked — collecting fresh signals.
star history
This repository releases pre-trained I3D (Inflated 3D ConvNet) models from the Quo Vadis paper for video classification. The models process both RGB and optical flow inputs and achieved state-of-the-art results on action recognition benchmarks including UCF101, HMDB51, and the Charades challenge. Built using TensorFlow and DeepMind’s Sonnet library.
Frequently asked
- What is google-deepmind/kinetics-i3d?
- Inflated 3D Convnet (I3D) CNN for video/action classification trained on the Kinetics dataset, released by DeepMind.
- Is kinetics-i3d open source?
- Yes — google-deepmind/kinetics-i3d is open source, released under the Apache-2.0 license.
- What language is kinetics-i3d written in?
- google-deepmind/kinetics-i3d is primarily written in Python.
- How popular is kinetics-i3d?
- google-deepmind/kinetics-i3d has 1.8k stars on GitHub.
- Where can I find kinetics-i3d?
- google-deepmind/kinetics-i3d is on GitHub at https://github.com/google-deepmind/kinetics-i3d.