search

Chiaraplizz / ST-TR

Spatial Temporal Transformer Network for Skeleton-Based Activity Recognition
MIT License
299 stars 58 forks source link

Understanding the Model Architecture #18

Open Nrohlable opened 3 years ago

Nrohlable commented 3 years ago

What will happen and what kind of results do we get if we have a window size of more than one and switch the model to spatial architecture ?

Chiaraplizz commented 3 years ago

Hi :)

The window size is set to 300 and temporal pooling is performed, until the final temporal dimension over the last layers is 75.

What do you mean by 'switching to spatial architecture'? Do you mean not considering the temporal relations?

Chiara

Nrohlable commented 3 years ago

Yes Chiara,

I meant if we don't considered the Temporal Relations and set the widow size explicitly to some value grater than 1 then in that case the prediction which comes as final result is on what basis ?