keras基于tensorflow后端---处理自己的数据

对于通道数为一的图像：

# -*- coding: utf-8 -*-
from __future__ import absolute_import
from __future__ import print_function
import os
from PIL import Image
import numpy as np
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.layers.advanced_activations import PReLU
from keras.layers.convolutional import Convolution2D, MaxPooling2D,Conv2D
from keras.optimizers import SGD, Adadelta, Adagrad
from keras.utils import np_utils, generic_utils
from six.moves import range

#数据预处理
#读取文件夹mnist下的42000张图片（灰度图）

def load_data():
    data = np.empty((2000,28,28,1),dtype="float32")      注意：：：： 这里2000是图像数，数目不对时，会报错
    label = np.empty((2000,),dtype="uint8")
   
    imgs = os.listdir("./mnist")
    num = len(imgs)
    for i in range(num):
        print(i)
        img = Image.open("./mnist/"+imgs[i])
        arr = np.asarray(img,dtype="float32")
       
        print(arr.shape)
        data[i,:,:,0] = arr
        #print(data.shape)
        label[i] = int(imgs[i].split('.')[0])
        #print(label[i])
    return data,label

def load_testData():
    testImg = Image.open('testNum1.jpg')
    imgArr = np.asarray(testImg,dtype="float32")
    testData = np.empty((1,28,28,1),dtype="float32")
    testData[0,:,:,0] = imgArr
    return testData

if __name__ == '__main__':

    testData = load_testData()
           
    #保存模型数据的文件
    filename = 'mnistCnnTrain.h5'
    #加载数据
    data, label = load_data()
   
    #print(data.shape[0], ' samples')
   
    #label为0~9共10个类别，转化格式
    label = np_utils.to_categorical(label, 10)
    #print(label)

    #=======================================
    #=======================================