The code about get the image features!

[haoopan]

Hi Cadene, Thank you open the source.But I face the problem about download the feature files, It's very slow to download,about 100 k/s. So can you share the code about extract image features. Thanks!

[Cadene]

@haoopan I am still working on releasing the code to extract image features to use pretrained models on new images / questions.

However I just uploaded the features on AWS. It should be faster.

mkdir -p data/vqa
cd data/vqa
wget http://data.lip6.fr/cadene/block/vqa2.tar.gz
#wget http://data.lip6.fr/cadene/block/coco.tar.gz
wget https://dl.fbaipublicfiles.com/cadene/block/coco.tar.gz
tar -xzvf vqa2.tar.gz
tar -xzvf coco.tar.gz

cd ../..
mkdir -p data/vqa/coco/extract_rcnn
cd data/vqa/coco/extract_rcnn
# wget http://data.lip6.fr/cadene/block/coco/extract_rcnn/2018-04-27_bottom-up-attention_fixed_36.tar
wget https://dl.fbaipublicfiles.com/cadene/block/extract_rcnn/2018-04-27_bottom-up-attention_fixed_36.tar
tar -xvf 2018-04-27_bottom-up-attention_fixed_36.tar

[tbozinis]

@Cadene What's the state of the image features extractor..? When do you think you'll release it? Thank you for your time.