About how to use classification (VehicleTypeNet)

I’d like to know how to adjust the input shape when I take the VehicleTypeNet to classification if my classification number is smaller than 6(I’m a rookie… I just thought make 6 dir and make some of dir null)

If you have 6 classes, please prepare 6 folders.
You can refer to the spec file in jupyter notebook.

I mean, this VehicleTypNet seems to be applicable to 6 categories, while I only have 3 categories to classify. What should I do to make him run? I tried to create three empty folders and made him run successfully

OK, so you are using the unpruned VehicleTypeNet model to train.
Refer to VehicleTypeNet | NVIDIA NGC and Image Classification — TAO Toolkit 3.21.11 documentation . Yes, you can create three empty folders for 3 untrained classes.

More, can you try another solution?

use_imagenet_head: False