length of sampled clips indicated in the paper

Hi,

After carefully read the paper, I'm confused about the clip length indicated in section 4 (Experiments) as follows:

> For the temporal domain, we randomly sample a clip (of alpha×T×tao frames) from the full-length video

where T is the number of frames sampled by the Slow pathway, tao is the temporal stride of the Slow pathway on the input frame (mentioned in section 3.1) and alpha is the frame rate ratio between the Fast and Slow pathway (mentioned in section 3.2)

So do the descriptions in section 5 (AVA experiments):

> The input is instantiation-specific alpha×T×tao frames of size 224×224.

> Inference. We perform inference on a single clip with alpha×T×tao frames around the frame that is to be evaluated.

Should the clips in the above quotes be just T×tao frames?

Thanks