尝试从图像预测数字时矩阵大小错误

时间:2019-02-16 15:13:28

标签: python-3.x keras prediction tf.keras

在编写第一个NN程序时遇到了一些麻烦。 我基本上想在画布上绘制一个数字,并将其提供给应该显示“已识别”数字的NN。

#create the NN, train with mnist and load img saved from canvas
def Analizza():

    mnist = tf.keras.datasets.mnist
    (x_train, y_train), (x_test, y_test) = mnist.load_data()

    x_train = tf.keras.utils.normalize(x_train, axis=1)

    model = tf.keras.models.Sequential()
    model.add(tf.keras.layers.Flatten())
    model.add(tf.keras.layers.Dense(128, activation=tf.nn.relu, input_shape= x_train.shape[1:]))
    model.add(tf.keras.layers.Dense(128, activation=tf.nn.relu))
    model.add(tf.keras.layers.Dense(10, activation=tf.nn.softmax))
    model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

    model.fit(x_train, y_train, epochs=3)
    model.save('modello_numeri.model')

    getter(c)

    width = 28
    height = 28

    img = Image.open(immagine_salvata)
    img = img.resize((width,height), Image.BICUBIC)
    img.save(immagine_salvata)

    img = PIL.Image.open("immagine.jpg").convert("L")
    img_ar = np.array(img)

    img_ar_norm = tf.keras.utils.normalize(img_ar, axis=1)

    plt.imshow(img_ar, cmap=plt.cm.binary)
    plt.show()  

    #make prediction
    predizione = model.predict(img_ar_norm, batch_size=None)
    risposta = np.argmax(predizione[0])

    print(risposta)

    #risultato = tk.Label(window, text=risposta, font=("Helvetica",57))
    #risultato.grid(row=0, column=1, sticky="E")

这部分给了我这些错误:

  

Tkinter回调中的异常   追溯(最近一次通话):    调用中的文件>“ C:\ Users \ Massimo \ AppData \ Local \ Programs \ Python \ Python37 \ lib \ tkinter__init __。py”,行1705      返回self.func(* args)    Analizza中的文件“ c:/Users/Massimo/AppVisualCode/disegno.py”,第91行       predizione = model.predict(img_ar_norm,batch_size = None)    预测中的文件“ C:\ Users \ Massimo \ AppData \ Roaming \ Python \ Python37 \ site-packages \ tensorflow \ python \ keras \ engine \ training.py”,行1113       自我,x,batch_size = batch_size,详细=详细,steps = steps)     文件“ C:\ Users \ Massimo \ AppData \ Roaming \ Python \ Python37 \ site-packages \ tensorflow \ python \ keras \ engine \ training_arrays.py”,行329,在model_iteration中       batch_outs = f(ins_batch)     调用中的文件“ C:\ Users \ Massimo \ AppData \ Roaming \ Python \ Python37 \ site-packages \ tensorflow \ python \ keras \ backend.py”,第3076行       run_metadata = self.run_metadata)     调用中的文件“ C:\ Users \ Massimo \ AppData \ Roaming \ Python \ Python37 \ site-packages \ tensorflow \ python \ client \ session.py”,行1439       run_metadata_ptr)     退出中的文件“ C:\ Users \ Massimo \ AppData \ Roaming \ Python \ Python37 \ site-packages \ tensorflow \ python \ framework \ errors_impl.py”,第528行       c_api.TF_GetCode(self.status.status))   tensorflow.python.framework.errors_impl.InvalidArgumentError:矩阵大小不兼容:In [0]:[28,28],In [1]:[784,128]            [[{{node MatMul}}]]

我知道预测所期望的矩阵大小应该有问题,但是为什么呢?我正在保存图像,将其调整为28x28(黑白),然后将其打开为np数组,就像它是x_train [0]

谢谢 最高

0 个答案:

没有答案