Prepare list file on TACC for imagenet

mxnet-prepare-imagenet.py

__author__ = 'zhengxu'

import os
import glob
import pickle


def make_record_io_list(image_net_folder, label_list_file, list_file):
    label_list = os.listdir(image_net_folder)
    with open(label_list_file, 'w') as f:
        pickle.dump(label_list, f)
    current_index = 0
    with open(list_file, 'w') as f:
        for i, label in enumerate(label_list):
            file_list = glob.glob(os.path.join(image_net_folder, label, '*.JPEG'))
            file_list_len = len(file_list)
            index_list = list(range(current_index, current_index + file_list_len))
            current_index += file_list_len
            string_list = zip(index_list, [i] * file_list_len, file_list)
            for str_tup in string_list:
                f.write("%d\t%d\t%s\r\n" % str_tup)

if __name__ == '__main__':
    TRAIN_DIR = '/work/03332/zhengxu/Data/ImageNet/ilsvrc12/train'
    VAL_DIR = '/work/03332/zhengxu/Data/ImageNet/ilsvrc12/val'
    # make_record_io_list(TRAIN_DIR, TRAIN_DIR+'/label_list.pickle', TRAIN_DIR+'/train.lst')
    make_record_io_list(VAL_DIR, VAL_DIR+'/label_list.pickle', VAL_DIR+'/train.lst')