Hi guys, where is the category file list for the models pre-trained in the ZOO?
We will need to upload that in order to use the model, right?
i’m not sure i understand your question
the list of category names that correspond to each of the output neurons of CNN. Even for ImageNet, they need to specified.
I think what is missing is a dictionary or so mapping the model output to something meaningful. I.e., right now, the network would return integer labels right? Say I use the model to do predictions on some (the same or different) dataset, how would I know what object an output of e.g., 23
corresponds to?
The indices of the dataset are in alphabetical order thanks to this line in the FolderDataset
.
Thus, we only need the map from the folder names to the classes, which can be found for example in this file, and is already in alphabetical order as well.
@fmassa thanks a lot, but does not work for me:
https://github.com/e-lab/pytorch-demos/tree/master/visor
I am not getting the right categories…
@Eugenio_Culurciello I believe the categories.txt
file does not contain the categories in the right order. The order should be the one present in the syncet_words
from here.
The alphabetical ordering I mentioned correspond to the syncet names
(n01440764 etc), and not the category names.