将图片转化为pickle或者mat数据格式

为了做小样本的实验,所以需要先将图片数据转化为pickle或者mat格式的数据,便于以后的读取,避免每次训练的时候,都要重新挨个图片读取在给标注。

这里我是看的tensorflow和Udacity合作的视频教程,所以学习到了将数据做乘pickle格式的方法

            try:
                with open(set_filename, 'wb') as f:
                    pickle.dump(dataset, f, pickle.HIGHEST_PROTOCOL)
            except Exception as e:
                print("无法制作 :", set_filename, e)

其主要步骤就是将你想要存储的数据存到“dataset"变量中,然后利用pickle.dump函数进行存储,一开始我使用480张512×512×3的图像存储的时候没有问题,但是,当数据量增加,便出现啦如下错误SystemError: error return without exception set

查找资料之后,根据https://blog.csdn.net/lj695242104/article/details/43062059

据说是pickl的一个bug,所以我换了一个路子,把数据存储为mat格式,经学习,发现python可以直接读写mat个数数据,而不需要使用matlab(省了不少事情啊),下面给出存储和读取mat格式数据的代码

首先是存储,我这里是将之前做好的两个pickle文件合并为了一个mat文件,之前合并车给为pickle的时候是会报错的,但是mat就没事。

同时也是用randomize函数将数据集的顺序打乱

# --*--coding:utf-8--*--
from __future__ import print_function
import imageio
import matplotlib.pyplot as plt
import numpy as np
import os
import sys
import tarfile
from IPython.display import display, Image
from sklearn.linear_model import LogisticRegression
from six.moves.urllib.request import urlretrieve
from six.moves import cPickle as pickle
import pprint
import scipy.io as sio


def randomize(dataset, labels):
  permutation = np.random.permutation(labels.shape[0])  # 根据labels的形状,获得一个随机的选取的顺序
  shuffled_dataset = dataset[permutation,:,:] # 然后根据这个顺序依次取出dataset中中的数据放到shuffled_dataset中
  shuffled_labels = labels[permutation] # 然后还是根据这个数据 将lable放到shuffled_labels中,
    # 这样就保证啦data与label的一一cuing挂希不改变
  return shuffled_dataset, shuffled_labels

folder = '你的路径'
class0_filename = '0.pickle'
class1_filename = '1.pickle'

pk1 = os.path.join(folder,class0_filename)
pk2 = os.path.join(folder,class1_filename)

path_list =[]
path_list.append(pk1)
path_list.append(pk2)

all_image = np.ndarray(shape=[960,512,512,3],dtype=np.float32)
all_label = np.ndarray(shape=[960],dtype=np.int32)
for index,pk in enumerate(path_list):


    pkl_file = open(pk, 'rb')

    images = pickle.load(pkl_file)
    label = np.ndarray(shape=[len(images)],dtype=np.int32)
    label[0:len(label)] = index
    all_image[index*480:(index+1)*480] = images
    all_label[index*480:(index+1)*480] = label

    # pprint.pprint(data1)
    print (pk,' 中images文件的形状是 ',images.shape)
    print (pk,' 中images[0]的形状是',images[0].shape)
    pkl_file.close()


print ('all_image shape is ',all_image.shape)
print ('all_label shape is ',all_label.shape)



shuffled_dataset,shuffled_labels = randomize(all_image,all_label)
print ('shuffed completed')
data = {
    'images':shuffled_dataset,
    'label':shuffled_labels
}

print ('begin to save to mat file')
sio.savemat('mattest.mat',data)
print ('mat file saved success')


然后是读取数据,这里读取完之后还进行啦几个图片的展示,确认以下数据是否正确。

# --*--coding:utf-8--*--

from __future__ import print_function
import imageio
import matplotlib.pyplot as plt
import numpy as np
import os
import sys
import tarfile
from IPython.display import display, Image
from sklearn.linear_model import LogisticRegression
from six.moves.urllib.request import urlretrieve
from six.moves import cPickle as pickle
import scipy.io as sio


pkl_file = open('datadata', 'rb')

# data = pickle.load(pkl_file)

data = sio.loadmat('mattest.mat')

print ('the keys in data is ', data.keys())

images = data['images']
label = data['label']

print ('image shape ',images.shape)
print ('label  shape ',label.shape)


# 遍历读取到的data和label,证实文件存储内容没有问题
for index ,image in enumerate(images):
    if index >475 and index <485:
        print (index,' image size is ',image.shape)
        print (index,' the label is ', label[0][index])
        plt.figure("class")
        plt.imshow(image)
        plt.show()

猜你喜欢

转载自blog.csdn.net/weixin_36193572/article/details/82110656