Is there a script on the training of VGG19 model using ImageNet and finding out the top-5 predictions?

[michaelTJC96]

From the paper published, I can see that an experiment was set up using VGG19 model to predict the classes for images from ImageNet. Thereafter, finding out the top-5 predictions. May I know whether there is such a script available?

Also, how do you conduct the attack on a large amount of images? From the code, there is a function “load_color_image_example_from_disk” which I am using to load the images from ImageNet dataset. However, do I have to load the images one by one?