Plans to load the bottom-up-attention detector weights or pointers on how to do so?

jwyang / graph-rcnn.pytorch

[ECCV 2018] Official code for "Graph R-CNN for Scene Graph Generation"

727 stars 158 forks source link

Plans to load the bottom-up-attention detector weights or pointers on how to do so? #84

Closed liunian-harold-li closed 4 years ago

liunian-harold-li commented 4 years ago

Hi, Jianwei, thank you for building this terrific repo!

I wonder if you have plans to load the bottom-up-attention detector weights (https://github.com/peteanderson80/bottom-up-attention) into this code?

If not, could you kindly provide some pointers on how one might do so?

Thanks!

liunian-harold-li commented 4 years ago

Upon further research, it seems it is almost impossible to do this as the models are implemented differently. It seems more realistic to just pre-train one with the attribute annotations from scratch.

Closing the issue.