Fetching the paper…

Knowing What, Where and When to Look: Efficient Video Action Modeling with Attention · Around