search

tgc1997 / RMN

IJCAI2020: Learning to Discretely Compose Reasoning Module Networks for Video Captioning
79 stars 12 forks source link

Could you upload your faster rcnn code to extract region features for my own data? #2

Closed AcodeC closed 4 years ago

daqingliu commented 4 years ago

We extracted region features by bottom-up-attention.

AcodeC commented 4 years ago

We extracted region features by bottom-up-attention.

Could you share the bottom-up-attention code which you have changed for extract MSVD or MSRVTT dataset? Thank you very much! Your work is awesome!

tgc1997 commented 4 years ago

Could you share the bottom-up-attention code which you have changed for extract MSVD or MSRVTT dataset? Thank you very much! Your work is awesome!

Our feature extracting code is stored in the server of our lab, but unfortunately, someone reinstalled the system without informing us so that the code was deleted. We extracted video region features based on extract_feature.py. This repo made some changes to the original repo and is easier to use, you could easily extract one image's region features by this code. As for MSVD and MSR-VTT, you can first convert a video to frames, and then extract features for the frames you want.

AcodeC commented 4 years ago

Could you share the bottom-up-attention code which you have changed for extract MSVD or MSRVTT dataset? Thank you very much! Your work is awesome!

Our feature extracting code is stored in the server of our lab, but unfortunately, someone reinstalled the system without informing us so that the code was deleted. We extracted video region features based on extract_feature.py. This repo made some changes to the original repo and is easier to use, you could easily extract one image's region features by this code. As for MSVD and MSR-VTT, you can first convert a video to frames, and then extract features for the frames you want.

That's a pity. And thank you very much!