zoukankan      html  css  js  c++  java
  • 深度学习 python 脚本实现 keras mninst 数字识别 预测端 code

    
    
    import numpy
    import skimage.io
    import matplotlib.pyplot as plt
    from keras.models import Sequential
    from keras.layers import Dense
    from keras.layers import Dropout
    from keras.layers import Flatten
    from keras.layers.convolutional import Conv2D
    from keras.layers.convolutional import MaxPooling2D
    from keras.models import load_model
    
    #if the picture is bigger than 28*28 will get below error
    #ValueError: cannot reshape array of size 775440 into shape (1,28,28,1)
    
    image = 'D:\sthself\ml\reshape7.jpg'
    
    img2 = skimage.io.imread(image,as_grey=True)
    skimage.io.imshow(img2)
    plt.show()
    
    #img3 is a matrix
    img3 = numpy.reshape(img2,(1,28,28,1)).astype('float32')
    
    print(img3)
    
    
    # rebuild the model  ,do we need to add the layer ?  AttributeError: 'Sequential' object has no attribute 'load_model'
    
    #If you stored the complete model, not only the weights, in the HDF5 file, then it is as simple as
    #from keras.models import load_model
    #model = load_model('model.h5')
    # examples https://stackoverflow.com/questions/35074549/how-to-load-a-model-from-an-hdf5-file-in-keras
    modelTrained = load_model('D:\works\jetBrians\PycharmProjects\tryPicture\my_model.h5')
    
    # we should get a correct answer is  2
    predict = modelTrained.predict(img3, verbose=0)
    #list of predicted labels and their probabilities
    print(predict[0])
    #[ 0.04785086  0.02547075  0.06954221  0.03620625  0.01439319  0.03016909   0.03120618  0.00815302  0.70513636  0.03187207]
    
    # AttributeError: 'Sequential' object has no attribute 'prediect_classes'
    result = modelTrained.predict_proba(img3,batch_size=1, verbose=0)
    print(result)
    
    print("tensorflow hello word is done")

    同事帮忙写的数字


    我自己写的数字



    程序打印log

    重点说明: 我们自己的图片应该是黑底白字 才能被识别

    D:applicationsAnaconda3python.exe D:/works/jetBrians/PycharmProjects/tryPicture/showPicture/ShowPicture.py
    Using TensorFlow backend.

    2018-03-08 20:43:29.102800: W C: f_jenkinshomeworkspace el-winMwindowsPY36 ensorflowcoreplatformcpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
    2018-03-08 20:43:29.102800: W C: f_jenkinshomeworkspace el-winMwindowsPY36 ensorflowcoreplatformcpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
    [  1.17482814e-05   1.08457927e-03   2.43252050e-02   3.06303948e-02
       1.07244858e-04   1.54377140e-05   1.01265108e-07   9.38272536e-01
       4.20123106e-04   5.13266213e-03]
    [[  1.17482814e-05   1.08457927e-03   2.43252050e-02   3.06303948e-02
        1.07244858e-04   1.54377140e-05   1.01265108e-07   9.38272536e-01
        4.20123106e-04   5.13266213e-03]]
    tensorflow hello word is done

    Process finished with exit code 0







  • 相关阅读:
    quiver()函数
    norm()函数
    求离散点的曲率
    WSGI接口简单介绍以及用 WSGI 协议的地方为何不直接用http?
    DLL编写中extern “C”和__stdcall的作用
    计算机中的单位
    Flask 源码阅读笔记 开篇
    一个Flask应用运行过程剖析
    Python 实例方法、类方法、静态方法的区别与作用以及私有变量定义和模块导入的区别
    flask之信号
  • 原文地址:https://www.cnblogs.com/TendToBigData/p/10501187.html
Copyright © 2011-2022 走看看