search

hengyuan-hu / bottom-up-attention-vqa

An efficient PyTorch implementation of the winning entry of the 2017 VQA Challenge.
GNU General Public License v3.0
754 stars 181 forks source link

is your model able to extract bottom-up attention features? #26

Closed SkylerZheng closed 6 years ago

SkylerZheng commented 6 years ago

Hi, I tried to re-implement Peter's bottom-up attention model (in caffe), but I failed. I'm wondering if your model can train on customized datasets and extract bottom-up attention features.

To me, your Readme is not clear. I don't know whether you can extract bottom-up features with your own model or not. It looks like you downloaded and applied bottom-up features from Peter's github repo.

hengyuan-hu commented 6 years ago

No the repo cannot extract features. To do that you need something like faster rcnn, which by itself is hard to implement. On Thu, Jun 21, 2018 at 12:20 PM Jian Zheng notifications@github.com wrote:

Hi, I tried to re-implement Peter's bottom-up attention model (in caffe), but I failed. I'm wondering if your model can train on customized datasets and extract bottom-up attention features.

To me, your Readme is not clear. I don't know whether you can extract bottom-up features with your own model or not. It looks like you downloaded and applied bottom-up features from Peter's github repo.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/hengyuan-hu/bottom-up-attention-vqa/issues/26, or mute the thread https://github.com/notifications/unsubscribe-auth/AEglZcud9r4vYYEW0b0FKxAiFf1ZS3EBks5t-_H4gaJpZM4UyoPJ .