Which neural network architecture could you use to classify videos?

Question $3$ options:

a max pooling layer that has no parameters

gated recurrent unit cell

pretrained Xception model

spatial patterns and depth wise patterns