python - 除非所有指定的输入大小都不为零，否则 Reshape 无法推断空张量的缺失输入大小

Question

我看了一些帖子，因为这似乎是一个非常常见的问题，但没有一个对我有用。我在一台装有 Tensorflow 1.11 的 Windows 机器上工作，但是当我尝试在一台装有 tensorflow 1.10 并更新到 1.13 的 Unbuntu 机器上时，它给了我这个错误

InvalidArgumentError (see above for traceback): Reshape cannot infer the missing input size for an empty tensor unless all specified input sizes are non-zero
 [[node Flatten_1/flatten/Reshape (defined at /home/matt/GitClone/PythonGitRepo/DNN/SelfDrivingCar/TrafficSignClassification/LeNETExample/Model/LeNet.py:69) ]]

我对问题所在有点迷茫，因为这应该是从 Windows 到 linux 的简单端口。

任何建议，将不胜感激。

这是程序的主要代码：

import pickle
import numpy as np
import matplotlib.pyplot as plt

from sklearn.utils import shuffle

import tensorflow as tf

from DataScripts import PreprocessData as predata
from Model import LeNet

# TODO: Fill this in based on where you saved the training and testing data

# Data set locations and paths
trainingPath = "TrainingData/train.p"
testingPath = "TrainingData/test.p"

# Get the data from the data sets using pickle to read them in
with open(trainingPath, mode="rb") as training_data:
    train = pickle.load(training_data)

with open(testingPath, mode="rb") as testing_data:
    test = pickle.load(testing_data)

# Create the features and labels for the data and link them
x_train, y_train = train['features'], train['labels']
x_test, y_test = test['features'], test['labels']

print("x_train shape:", x_train.shape)
print("y_train shape:", y_train.shape)
print("x_test shape:", x_test.shape)
print("y_test shape:", y_test.shape)


# Define number of training samples and number of testing samples
num_train = len(x_train)
num_test = len(x_test)
# Define the number of classes (only the unique vales in y sets)
num_classes = len(np.unique(y_train))

# Define the image shape and data shape
image_shape = x_train[0].shape
print("Num of Training Samples:{} - Num of Test Sample:{} - Num Of classes:{}".format(num_train, num_test, num_classes))

# histogram of label frequency Simple data exploration
hist, bins = np.histogram(y_train, bins=num_classes)
width = 0.7 * (bins[1] - bins[0])
center = (bins[:-1] + bins[1:]) / 2
plt.bar(center, hist, align='center', width=width)
plt.show()


# Process all the data and create validation data

x_train, x_test_normalized, x_validation, y_validation =     predata.preprocessData(x_train, y_train, x_test)


# Build LeNET Model
EPOCHS = 60
BATCH_SIZE = 100
rate = 0.0009 # Learning rate

tf.reset_default_graph()
# Tensor placehoders
x = tf.placeholder(tf.float32, (None, 32, 32, 1)) # Define shape of a image 32X32
y = tf.placeholder(tf.int32, (None))              # Class label
keep_prob = tf.placeholder(tf.float32)            # probability to keep     units
one_hot_y = tf.one_hot(y, 43)


# Start the training

logits = LeNet.makeModel(x, keep_prob)
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(logits = logits, labels=one_hot_y)
loss_operation = tf.reduce_mean(cross_entropy)
optimizer = tf.train.AdamOptimizer(learning_rate = rate)
training_operation = optimizer.minimize(loss_operation)

correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(one_hot_y, 1))
accuracy_operation = tf.reduce_mean(tf.cast(correct_prediction,     tf.float32))
saver = tf.train.Saver()

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    num_examples = len(x_train)

    print("Training...")
    print()
    for i in range(EPOCHS):
        x_train, y_train = shuffle(x_validation, y_validation)
        for offset in range(0, num_examples, BATCH_SIZE):
        end = offset + BATCH_SIZE
        batch_x, batch_y = x_train[offset:end], y_train[offset:end]
            sess.run(training_operation, feed_dict={x: batch_x, y: batch_y, keep_prob: 0.5})

        validation_accuracy = LeNet.evaluate(x_validation, y_validation, BATCH_SIZE, accuracy_operation, keep_prob, x, y)
        print("EPOCH {} ...".format(i + 1))
        print("Validation Accuracy = {:.3f}".format(validation_accuracy))
        print()

    saver.save(sess, '.\lenet')
    print("Model saved")


with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    saver2 = tf.train.import_meta_graph('./lenet.meta')
    saver2.restore(sess, "./lenet")
    # Test accuracy on test sets
    test_accuracy = LeNet.evaluate(x_test_normalized, y_test, BATCH_SIZE, accuracy_operation, keep_prob, x, y)
    print("Test Set Accuracy = {:.3f}".format(test_accuracy))

这是 LeNEt 方法：

import tensorflow as tf
#from tensorflow.contrib.layers import flatten

def makeModel(x, keep_prob):
    mu = 0
    sigma = 0.1

    # Layer 1: Convolutional. Input = 32x32x1. Output = 28x28x6.
    W1 = tf.Variable(tf.truncated_normal(shape=(5, 5, 1, 6), mean=mu, stddev=sigma), name="W1")
    x = tf.nn.conv2d(x, W1, strides=[1, 1, 1, 1], padding='VALID')
    b1 = tf.Variable(tf.zeros(6), name="b1")
    x = tf.nn.bias_add(x, b1)
    print("layer 1 shape:", x.get_shape())

    # Activation function.
    x = tf.nn.relu(x)

    # Max Pooling. Input = 28x28x6. Output = 14x14x6.
    x = tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
    layer1 = x

    # Layer 2: Convolutional. Output = 10x10x16.
    W2 = tf.Variable(tf.truncated_normal(shape=(5, 5, 6, 16), mean=mu, stddev=sigma), name="W2")
    x = tf.nn.conv2d(x, W2, strides=[1, 1, 1, 1], padding='VALID')
    b2 = tf.Variable(tf.zeros(16), name="b2")
    x = tf.nn.bias_add(x, b2)

    # Activation function.
    x = tf.nn.relu(x)

    # Max Pooling. Input = 10x10x16. Output = 5x5x16.
    x = tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')
    layer2 = x

    # Layer 3: Convolutional. Output = 1x1x400.
    W3 = tf.Variable(tf.truncated_normal(shape=(5, 5, 16, 400), mean=mu, stddev=sigma), name="W3")
    x = tf.nn.conv2d(x, W3, strides=[1, 1, 1, 1], padding='VALID')
    b3 = tf.Variable(tf.zeros(400), name="b3")
    x = tf.nn.bias_add(x, b3)

    # Activation function.
    x = tf.nn.relu(x)
    layer3 = x

    # Flatten. Input = 5x5x16. Output = 400.
    layer2flat = tf.layers.flatten(layer2)
    print("layer2flat shape:", layer2flat.get_shape())

    # Flatten x. Input = 1x1x400. Output = 400.
    xflat = tf.layers.flatten(x)
    print("xflat shape:", xflat.get_shape())

    # Concat layer2flat and x. Input = 400 + 400. Output = 800
    x = tf.concat([xflat, layer2flat], 1)
    print("x shape:", x.get_shape())

    # Dropout
    x = tf.nn.dropout(x, keep_prob)

    # Layer 4: Fully Connected. Input = 800. Output = 43.
    W4 = tf.Variable(tf.truncated_normal(shape=(800, 43), mean=mu, stddev=sigma), name="W4")
    b4 = tf.Variable(tf.zeros(43), name="b4")
    logits = tf.add(tf.matmul(x, W4), b4)

    return logits

# x and y are tensorflow place holders
def evaluate(x_validation, y_validation, BATCH_SIZE, accuracy_operation, keep_prob, x, y):
    num_examples = len(x_validation)
    total_accuracy = 0
    sess = tf.get_default_session()
    for offset in range(0, num_examples, BATCH_SIZE):
        batch_x, batch_y = x_validation[offset:offset+BATCH_SIZE], y_validation[offset:offset+BATCH_SIZE]
        accuracy = sess.run(accuracy_operation, feed_dict={x: batch_x, y: batch_y, keep_prob: 1.0})
        total_accuracy += (accuracy * len(batch_x))
    return total_accuracy / num_examples

这是数据的预处理代码

from sklearn.utils import shuffle
from sklearn.model_selection import train_test_split
import numpy as np

def preprocessData(x_train, y_train, x_test):
    # Convert to grayscale
    x_train_rgb = x_train
    x_train_gry = np.sum(x_train / 3, axis=3, keepdims=True)

    x_test_rgb = x_test
    x_test_gry = np.sum(x_test / 3, axis=3, keepdims=True)

    print('RGB:', x_train_rgb.shape)
    print('Grayscale:', x_train_gry.shape)

    x_train = x_train_gry
    x_test = x_test_gry

    ## Normalize the train and test datasets to (-1,1)

    x_train_normalized = (x_train - 128) / 128
    x_test_normalized = (x_test - 128) / 128

    ## Shuffle the training dataset and create the validation set
    # Validatioin set will not be used by Training

    x_train_normalized, y_train = shuffle(x_train_normalized, y_train)
    x_train, x_validation, y_train, y_validation = train_test_split(x_train_normalized, y_train,
                                                                    test_size=0.20, random_state=42)

    print("Old X_train size:", len(x_train_normalized))
    print("New X_train size:", len(x_train))
    print("X_validation size:", len(x_validation))
    return x_train, x_test_normalized, x_validation, y_validation

数据集可以在这里找到 https://d17h27t6h515a5.cloudfront.net/topher/2017/February/5898cd6f_traffic-signs-data/traffic-signs-data.zip

标签： pythontensorflowcomputer-vision