The imagenet.py
provides a basic dataset which loads images from a given .txt
file. This file usually contains the directory path of images like
n01440764/n01440764_10026.JPEG
n01440764/n01440764_10027.JPEG
n01440764/n01440764_10029.JPEG
n01440764/n01440764_10040.JPEG
n01440764/n01440764_10042.JPEG
n01440764/n01440764_10043.JPEG
n01440764/n01440764_10048.JPEG
Such that, by modifing data_root
and ann_file
in config file, we could launch the pretraining on any other custom dataset.