Could you please provide the pre computed head_feats and ann_feats

[BigRedT]

Running mask-rcnn to compute these features takes some time. It would be great if you could provide these features to download and use directly.

Thanks!

[lichengunc]

Hi, the features are too big to share. I extracted image's C4 features first and saved them there. During training, I run ROI-pooling for each object given ground-truth or detected bounding box. Specifically, for subject module, I will resize the object's roi-pooled feature to 7x7x2048, to be attended by language embedding. Hope this explanation could help you. :-)