search

GriffinLiang / vrd-dsr

Code for Visual Relationship Detection with Deep Structural Ranking (AAAI2018)
121 stars 32 forks source link

Question on gt.mat or gt.pkl #7

Closed yangshao closed 6 years ago

yangshao commented 6 years ago

It seems that all the information required has already been incorporated into the train.pkl/test.pkl, am i right?

GriffinLiang commented 6 years ago

Yes. These two files contain all the annotation information of VRD dataset.

yangshao commented 6 years ago

@GriffinLiang Thanks, another related question is why do you use Recall@K to evaluate the predicate prediction task, I feel it's a pure classification problem, so why not using accuracy?

GriffinLiang commented 6 years ago

The original eccv2016 paper suggests to use recall due to the incompleteness of annotations.

2018-04-19 15:19 GMT+08:00 yangshao notifications@github.com:

@GriffinLiang https://github.com/GriffinLiang Thanks, another related question is why do you use Recall@K to evaluate the predicate prediction task, I feel it's a pure classification problem, so why not using accuracy?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GriffinLiang/vrd-dsr/issues/7#issuecomment-382636023, or mute the thread https://github.com/notifications/unsubscribe-auth/AGupmqxETqdNzp3d-w3onSNpEwfl5BMuks5tqDqGgaJpZM4TbAlz .