The imagenet directory structure is laid out as follows Top-level dirs: train, test, valid
(you should provide path to those dirs as input)
Mid-level dirs: class1, class2, ...
One directory for each class. The folders directly contain the images.

df_from_imagenet_dir(dirs)

Arguments

dirs

character
List of dirs to create dataframes from.

Value

data.frame
with columns "image" and "class" (the class).