使用 TensorFlow 进行图像分类，值错误

Question

我是深度学习和 tensorflow 的新手，我正在尝试使用 tensorflow 创建一个图像分类器，它将对 5 类图像进行分类。我的训练数据集是 25000 张图像，测试数据集是 5000 张图像。下面是我的代码：

import os
from random import shuffle

import cv2
import matplotlib.pyplot as plt
import numpy as np
import tensorflow as tf
import tflearn
from tflearn.layers.conv import conv_2d, max_pool_2d
from tflearn.layers.core import input_data, dropout, fully_connected
from tflearn.layers.estimator import regression
from tqdm import tqdm
import keras
from keras.models import save_model
"""from keras.models import Sequential
from keras.layers import Dense"""

TRAIN_DIR = 'train'
TEST_DIR = 'test'
IMG_SIZE = 16
LR = 1e-3
MODEL_NAME = 'cifar 10 -convnet'

def create_label(image_name):
    """ Create an one-hot encoded vector from image name """
    word_label = image_name.split('.')[0:2]
    if word_label == 'cat':
        return np.array([1,0,0,0,0])
    elif word_label == 'Dog':
        return np.array([0,1,0,0,0])
    elif word_label == 'Automobile':
        return np.array([0,0,1,0,0])
    elif word_label == 'Airplane':
        return np.array([0,0,0,1,0])
    elif word_label == 'Ship':
        return np.array([0,0,0,0,1])


def create_train_data():
    training_data = []
    for img in tqdm(os.listdir(TRAIN_DIR)):
        path = os.path.join(TRAIN_DIR, img)
        img_data = cv2.imread(path, cv2.IMREAD_GRAYSCALE)
        img_data = cv2.resize(img_data, (IMG_SIZE, IMG_SIZE))
        training_data.append([np.array(img_data), create_label(img)])
    shuffle(training_data)
    np.save('train_data.npy', training_data)
    return training_data


def create_test_data():
    testing_data = []
    for img in tqdm(os.listdir(TEST_DIR)):
        path = os.path.join(TEST_DIR, img)
        img_num = img.split('.')[0:2]
        img_data = cv2.imread(path, cv2.IMREAD_GRAYSCALE)
        img_data = cv2.resize(img_data, (IMG_SIZE, IMG_SIZE))
        testing_data.append([np.array(img_data), img_num])

    shuffle(testing_data)
    np.save('test_data.npy', testing_data)
    return testing_data


# If dataset is not created:
train_data = create_train_data()
test_data = create_test_data()
# If you have already created the dataset:
# train_data = np.load('train_data.npy')
# test_data = np.load('test_data.npy')
train = train_data[:25000]
test = train_data[:5000]
X_train = np.array([i[0] for i in train]).reshape(-1, IMG_SIZE, IMG_SIZE, 1)
y_train = [i[1] for i in train]
X_test = np.array([i[0] for i in test]).reshape(-1, IMG_SIZE, IMG_SIZE, 1)
y_test = [i[1] for i in test]

# Building The Model

tf.reset_default_graph()
convnet = input_data(shape=[None, IMG_SIZE, IMG_SIZE, 1], name='input')
convnet = conv_2d(convnet, 32, 5, activation='relu')
convnet = max_pool_2d(convnet, 5)
convnet = conv_2d(convnet, 64, 5, activation='relu')
convnet = max_pool_2d(convnet, 5)
convnet = conv_2d(convnet, 128, 5, activation='relu')
convnet = max_pool_2d(convnet, 5)
convnet = conv_2d(convnet, 64, 5, activation='relu')
convnet = max_pool_2d(convnet, 5)
convnet = conv_2d(convnet, 32, 5, activation='relu')
convnet = max_pool_2d(convnet, 5)
convnet = fully_connected(convnet, 1024, activation='relu')
convnet = dropout(convnet, 0.8)
convnet = fully_connected(convnet, 2, activation='softmax')
convnet = regression(convnet, optimizer='adam', learning_rate=LR, loss='categorical_crossentropy', name='targets')
model = tflearn.DNN(convnet, tensorboard_dir='log', tensorboard_verbose=0)
history = model.fit({'input': X_train}, {'targets': y_train}, n_epoch=25,
          validation_set=({'input': X_test}, {'targets': y_test}),
          snapshot_step=500, show_metric=True, run_id=MODEL_NAME)

#print(history.history.keys())
#plt.figure(1)

# summarize history for accuracy

"""plt.subplot(211)
plt.plot(history.history['acc'])
plt.plot(history.history['val_acc'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test1'], loc='upper left')
plt.show()"""

fig = plt.figure(figsize=(25, 12))

for num, data in enumerate(test_data[:25]):

    img_num = data[1]
    img_data = data[0]

    y = fig.add_subplot(5, 5, num + 1)
    orig = img_data
    data = img_data.reshape(IMG_SIZE, IMG_SIZE, 1)
    model_out = model.predict([data])[0]

    if np.argmax(model_out) == 1:
        str_label = 'Dog'
    if np.argmax(model_out) == 2:
        str_label = 'Automobile'
    if np.argmax(model_out) == 3:
        str_label = 'Airplane'
    if np.argmax(model_out) == 4:
        str_label = 'Ship'
    else:
        str_label = 'Cat'

    y.imshow(orig, cmap='gray')
    plt.title(str_label)
    y.axes.get_xaxis().set_visible(False)
    y.axes.get_yaxis().set_visible(False)
plt.show()

我收到以下错误： ValueError：无法为 Tensor 'targets/Y:0' 提供形状 (64,) 的值，其形状为“(?, 2)”

有人可以帮我解决这个问题吗？提前致谢。

Answer 1

问题出在标签的形状上，因此请检查 y_train 和 y_test 的形状。它们都必须具有类似于模型输出的形状，显示在错误 ...which has shape '(?, 2)' 处，并在您创建最后一个完全连接层 convnet = fully_connected(convnet, 2, activation='softmax') 的行处定义。

使用 TensorFlow 进行图像分类，值错误

Image Classification Using TensorFlow, Value error

python

conv-neural-network

tensorflow

tflearn