cdancette / rubi.bootstrap.pytorch

NeurIPS 2019 Paper: RUBi : Reducing Unimodal Biases for Visual Question Answering

how to extract image features

ProbeTS opened this issue 3 years ago · comments

ProbeTS commented 3 years ago

Could you please share the code to extract image features "2018-04-27_bottom-up-attention_fixed_36" or the pre-trained RCNN?

Corentin Dancette commented 3 years ago

You can download features here: wget http://data.lip6.fr/cadene/block/coco/extract_rcnn/2018-04-27_bottom-up-attention_fixed_36.tar

You can extract them using https://gitlab.com/vedanuj/vqa-maskrcnn-benchmark