手把手创建自己的数据集:类似cifar10数据结构的数据集

前言：我们在跑神经网络时，通常使用的都是别人已经整理好的数据集，如MNIST、CIFAR10、CIFAR100等，但是在实际的应用中，往往需要根据实际的问题创建对应该问题的数据集，这就需要用自己的图片创建一个类似CIFAR10的数据集。如果我们创建的数据集格式和CIFAR10的格式一样的，那么所创建的数据集将很容易地输入到原有的神经网络，而无需改动太多的结构。

1、首先查看CIFAR10数据集是长什么样子的？

CIFAR10经过解压后会得到cifar-100batches-py的文件夹，如下图所示：

通过dataset2Image.py文件可以将以上的数据batch解压成图片?

import pickle

def unpickle(file):
  with open(file, 'rb') as fo:
    #dict = cPickle.load(fo)
    dict = pickle.load(fo, encoding='iso-8859-1')  # encoding='bytes'
    #dict = pickle.load(fo, encoding='bytes')
  return dict

if __name__ == '__main__':
  #print (unpickle('train_batch_video'))
  print(unpickle('data_batch_1'))

?输出的结果如下：

因此，我们只需要生成类似的数据格式即可。?

2、工程文件结构

?

?保存数据的文件夹在data目录下：

1、batch_save_train: 用于保存训练集的batch；

2、batch_save_val: 用于保存验证集或者测试集的batch；

3、figure_name_label_train: 用于保存生成的训练集图片名和标签的txt文件；

4、figure_name_label_val: 用于保存生成的验证集或者测试集图片名和标签的txt文件；

5、train: 训练集数据，里面包含了以分好类的数据；

?image2dataset.py （代码用电脑端看比较好）

# -*- coding: UTF-8 -*-

import cv2
import os
import numpy as np


DATA_LEN = 3072     # 32x32x3=3072
#DATA_LEN = 43200     # 160x90x3
CHANNEL_LEN = 1024  # 32x32=1024
#CHANNEL_LEN = 14400  # 160x90 = 14400
SHAPE = (32, 32)#(160, 90)#32

# 修改
#figure_path = '/home/user/PycharmProjects/DataSet_ipanel/Image2Dataset/layoutdata-160-90/train/video'
#figure_name_label = '/home/user/PycharmProjects/DataSet_ipanel/Image2Dataset/layoutdata-160-90/figure_name_label_train/image_train_video_list.txt'
#batch_save = '/home/user/PycharmProjects/DataSet_ipanel/Image2Dataset/layoutdata-160-90/batch_save_train'
## 修改imagelist()标签值

figure_path = './data/train/airbus'  # 图片的位置
figure_name_label = './data/figure_name_label_train/image_train_airbus_list.txt' # 保存图片名称和标签
batch_save = './data/batch_save_train'  # 保存batch文件
## 修改imagelist()标签值

def imread(im_path, shape=None, color="RGB", mode=cv2.IMREAD_UNCHANGED):
    im = cv2.imread(im_path, cv2.IMREAD_UNCHANGED)
    if color == "RGB":
        im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB)
    if shape != None:
        #assert isinstance(shape, int)
        #im = cv2.resize(im, (shape, shape))
        im = cv2.resize(im, shape)
    return im


def read_data(filename, data_path, shape=None, color='RGB'):
    """
       filename (str): a file
         data file is stored in such format:
           image_name  label
       data_path (str): image data folder
       return (numpy): a array of image and a array of label
    """
    (shape1, shape2) = shape
    if os.path.isdir(filename):
        print("Can't found data file!")
    else:
        f = open(filename)
        lines = f.read().splitlines()
        count = len(lines)
        data = np.zeros((count, DATA_LEN), dtype=np.uint8)
        # label = np.zeros(count, dtype=np.uint8)
        lst = [ln.split(' ')[0] for ln in lines]
        label = [int(ln.split(' ')[1]) for ln in lines]

        idx = 0
        #s, c = SHAPE, CHANNEL_LEN
        c = CHANNEL_LEN
        for ln in lines:
            fname, lab = ln.split(' ')
            #im = imread(os.path.join(data_path, fname), shape=s, color='RGB')
            im = imread(os.path.join(data_path, fname), shape=SHAPE, color='RGB')
            '''
            im = cv2.imread(os.path.join(data_path, fname), cv2.IMREAD_UNCHANGED)
            im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB)
            im = cv2.resize(im, (s, s))
            '''
            data[idx, :c] = np.reshape(im[:, :, 0], c)
            data[idx, c:2 * c] = np.reshape(im[:, :, 1], c)
            data[idx, 2 * c:] = np.reshape(im[:, :, 2], c)
            label[idx] = int(lab)
            idx = idx + 1

        return data, label, lst


def py2bin(data, label):
    label_arr = np.array(label).reshape(len(label), 1)
    label_uint8 = label_arr.astype(np.uint8)
    arr = np.hstack((label_uint8, data))

    with open(batch_save, 'wb') as f: # 每个文件夹修改
    #with open('/home/user/PycharmProjects/DataSet_ipanel/layoutdata-160-90/train/train_batch/train_batch_big5small5', 'wb') as f:
        for element in arr.flat:
            f.write(element)

import pickle
BIN_COUNTS = 1  # 每一类的数据为一个batch
def pickled(savepath, data, label, fnames, bin_num=BIN_COUNTS, mode="train", name=None):
    '''
      savepath (str): save path
      data (array): image data, a nx3072 array
      label (list): image label, a list with length n
      fnames (str list): image names, a list with length n
      bin_num (int): save data in several files
      mode (str): {'train', 'test'}
    '''
    assert os.path.isdir(savepath)
    total_num = len(fnames)
    samples_per_bin = total_num // bin_num  # 将/换为// （TypeError: slice indices must be integers or None or have an __index__ method）
    assert samples_per_bin > 0
    idx = 0
    for i in range(bin_num):
        start = i * samples_per_bin
        end = (i + 1) * samples_per_bin

        if end <= total_num:
            dict = {'data': data[start:end, :],
                    'labels': label[start:end],
                    'filenames': fnames[start:end]}
        else:
            dict = {'data': data[start:, :],
                    'labels': label[start:],
                    'filenames': fnames[start:]}
        if mode == "train":
            dict['batch_label'] = "training batch {}".format(name)#(idx, bin_num)
        else:
            dict['batch_label'] = "testing batch {}".format(name)#(idx, bin_num)

        with open(os.path.join(savepath, 'data_batch_' + str(name)), 'wb') as fi:#str(idx)), 'wb') as fi:
            # cPickle.dump(dict, fi)
            pickle.dump(dict, fi)
        #idx = idx + 1

def imagelist():
    directory_normal = figure_path
    #directory_normal = r"/home/user/PycharmProjects/DataSet_ipanel/layoutdata-160-90/train/big5small6"  # 原始图片位置，32*32 pixel
    file_train_list = figure_name_label
    #file_train_list = r"/home/user/PycharmProjects/DataSet_ipanel/layoutdata-160-90/train/image_train_big5small6_list.txt"  # 构建imagelist输出位置
    with open(file_train_list, "a") as f:
        for filename in os.listdir(directory_normal):
            #f.write(filename + " " + "0" + "\n")    #这里分类默认全为0
            f.write(filename + " " + "0" + "\n")  # 这里分类默认全为0 ##########


if __name__ == '__main__':
    data_path = figure_path
    #data_path = '/home/user/PycharmProjects/DataSet_ipanel/layoutdata-160-90/train/big5small6'
    file_list = figure_name_label
    #file_list = '/home/user/PycharmProjects/DataSet_ipanel/layoutdata-160-90/train/image_train_big5small6_list.txt'
    save_path = batch_save#'./bin'
    imagelist() #构建imagelist  # 生成名字和标签的对应关系
    data, label, lst = read_data(file_list, data_path, shape=SHAPE)    #将图片像素数据转成矩阵和标签列表
    #py2bin(data, label) #将像素矩阵和标签列表转成cifar10 binary version # 二进制版本
    pickled(save_path, data, label, lst, bin_num=1, name='airbus')  # 生成python版本

还是截图吧。。。

生成了data_batch_airbus文件
?

对data_batch_airbus文件进行提取，如下所示：?

??

可以的到和之前cifar10解压的数据一样的格式，这种做法是每个batch都是一类图片，在训练时对训练数据进行随机打乱即可。（具体可以到我主页（我的博客）查看代码）。文字感觉还是很难表达清楚，需要的同学看代码自己理解下就OK了，并不是很复杂，有需要的同学可以看下收藏、转发，不懂的同学可以留言@我，有时间会尽量解答。