gina9726/deep-rtc

Codes for "Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier"

deep-rtc's Issues

Question regarding CIFAR 100 preparation

Hi, I have a question regarding CIFAR 100 preparation.
Usually, we utilize CIFAR 100 as a python-version on their webpage, which is divided into test and train files.
However, according to the data preparation section on README.md, it should be saved as jpg with class hierarchy. According to train/valid/test splits in https://github.com/gina9726/Deep-RTC/tree/master/prepro/splits/cifar100, the file format is "Image{:5d}.jpg" and some indices are missing.

Could you tell me how to form CIFAR 100 dataset?

The Long-tailed version link just denotes the various long-tailed versions of tfrecord files.

Thank you.