cifar100的获取方式以及处理方式

GUO_PP

于 2024-09-18 11:15:18 发布

阅读量151

点赞数 2

文章标签：目标检测

本文链接：https://blog.csdn.net/guo_pp/article/details/142328337

版权

注：最终图片数据集以 100 分类文件夹存放。
CSDN 站内下载转换后数据集：CIFAR-10、CIFAR-100
Github：
CIFAR-10 dataset by classes folder
CIFAR-100 dataset by classes folder

cifar100的转换代码：

from PIL import Image
import numpy as np
import pickle
import os
from tqdm import trange
from os.path import join

def my_mkdirs(path):

if not os.path.exists(path):
    os.makedirs(path)

def unpickle(file):

with open(file, 'rb') as fo:
    dict = pickle.load(fo, encoding='latin1')

return dict

settings

src_dir = ‘/home/gpc/disk_1/datasets/cifar-100-python’ # the dir you uncompress the dataset
dst_dir = ‘/home/gpc/disk_1/datasets/cifar100’ # the dir you want the img_dataset to be

if name == ‘main’:
meta = unpickle(join(src_dir, ‘meta’)) # KEYS: {‘fine_label_names’, ‘coarse_label_names’}
my_mkdirs(dst_dir)

for data_set in ['train', 'test']:
    print('Unpickling {} dataset......'.format(data_set))
    data_dict = unpickle(join(src_dir, data_set)) # KEYS: {'filenames', 'batch_label', 'fine_labels', 'coarse_labels', 'data'}
    my_mkdirs(join(dst_dir, data_set))

    for fine_label_name in meta['fine_label_names']:
        my_mkdirs(join(dst_dir, data_set, fine_label_name))

    for i in trange(data_dict['data'].shape[0]):
        img = np.reshape(data_dict['data'][i], (3, 32, 32))
        i0 = Image.fromarray(img[0])
        i1 = Image.fromarray(img[1])
        i2 = Image.fromarray(img[2])
        img = Image.merge('RGB', (i0, i1, i2))
        img.save(join(dst_dir, data_set, meta['fine_label_names'][data_dict['fine_labels'][i]], data_dict['filenames'][i]))

print('All done.')

转载：https://blog.csdn.net/weixin_43667077/article/details/108838343