search

MILVLG / bottom-up-attention.pytorch

A PyTorch reimplementation of bottom-up-attention models
Apache License 2.0
294 stars 76 forks source link

some questions about num_classes #80

Closed Zhongan-Wang closed 2 years ago

Zhongan-Wang commented 3 years ago

Thanks for your code.When I try to train the vg datasets by myself, i find the vg datasets num_classes is 1601 and I must set 1601 make sure the code work,but I find you are using 401,in my code ,this will appear some problem.I don't know why you set 401.

MIL-VLG commented 3 years ago

400 is the number of attributes and 1600 is the number of object classes