facebookresearch / facebookresearch/SlowFast

Release of 32-frame models? (16x8, etc)

Open
#125 0 comments 0 reactions 0 assignees View on GitHub
enhancement
Dominant language
Python
Stars
7.4k
Forks
1.3k
PR merge metrics
No merged PRs in 30d

Description

Hi, I've really enjoyed adapting your model to my tasks. Thank you very much for the release.

I notice that months later, still only "8x8" checkpoints are released for Kinetics, with "16x8" model marked as "coming soon" in the link. Given that "16x8" model are available for AVA, would it be easy to link the checkpoints for Kinetics please?

I am working on a task that may benefit from more frames. As you know the "8x8" model is really using 32 frames, downsampled to 16 frames. Very curious if finetuning on longer sequences helps a lot.

Could you please link the Kinetics model, so it probably already exists and not a secret?

Thank you!

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.