The goal of the Kinetics dataset is to help the computer vision and machine learning communities advance models for video understanding. Given this large human action classification dataset, it may be possible to learn powerful video representations that transfer to different video tasks.
: How approachable and welcoming the hosts are to those practicing naturism for the first time.
Your report may want to distinguish between these two concepts:
: How approachable and welcoming the hosts are to those practicing naturism for the first time.
Your report may want to distinguish between these two concepts:
1. Possible to use ImageNet checkpoints?
We allow finetuning from public ImageNet checkpoints for the supervised track -- but a link to the specific checkpoint should be provided with each submission.
2. Possible to use optical flow?
Flow can be used as long as not trained on external datasets, except if they are synthetic.
naturist freedom at monikas home top
3. Can we train on test data without labels (e.g. transductive)?
No.
: How approachable and welcoming the hosts are
4. Can we use semantic class label information?
Yes, for the supervised track.
naturist freedom at monikas home top
5. Will there be special tracks for methods using fewer FLOPs / small models or just RGB vs RGB+Audio in the self-supervised track?
We will ask participants to provide the total number of model parameters and the modalities used and plan to create special mentions for those doing well in each setting, but not specific tracks.