rohitgirdhar/AttentionalPoolingAction

bottom up saliency

nationalflag opened this issue · 2 comments

So, the algorithm seems to replace the average pooling with a weighted pooling using bottom up saliency as the weight?

The bottom up saliency maps showed in Fig 2 are pose regularized or not?

Reg the algorithm: yes that's correct
Reg saliency maps in Fig 2: They are not pose-regularized.