Kinetics 400 models

[lovelyczli]

Thanks for your Pytorch implementation for MoViNet.
Could you please provide a model trained on Kinetics 400 as Table 9 in [1].
It is quite important for our future works.
Looking forward to your reply.

[1] MoViNets: Mobile Video Networks for Efficient Video Recognition

[Hyperparticle]

Hi, the official repo now has Kinetics 400 pretrained checkpoints for base models available here (streaming models should be available soon): https://github.com/tensorflow/models/tree/master/official/projects/movinet#kinetics-400. Accuracy should be a bit better than what is reported in the paper. You should be able to convert this to PyTorch similarly to the Kinetics 600 models.