Train on custom dataset
KleinXin opened this issue · comments
HotelID is a dataset that has similar hierarchical architecture as Imagenet21k.
https://arxiv.org/abs/2106.05746
So I want to try to use this project to train on that dataset.
The question is in that case, each chain_id has many sub-hotels, this means the parent of the chain_id class is itself when genrating the 'semantic tree'.
Do you think it can give reasonable results?
your tree doesn't have lots of depth (only 2 possible hierarchies if i understand correctly).
so a simple 2 softmax solution seems reasonable, and might give additional benefit compared to just predicting the sub-class (chain_id)