我想输入批量的混洗图像进行训练，然后根据the generic input images in TensorVision编写代码，但是我收到错误。我无法弄清楚错误的地方。 这是我的代码：
import os
import tensorflow as tf
def read_labeled_image_list(image_list_file):
"""
Read a .txt file containing pathes and labeles.
Parameters
----------
image_list_file : a .txt file with one /path/to/image per line
label : optionally, if set label will be pasted after each line
Returns
-------
List with all filenames in file image_list_file
"""
f = open(image_list_file, 'r')
filenames = []
labels = []
for line in f:
filename, label = line[:-1].split(' ')
filenames.append(filename)
labels.append(int(label))
return filenames, labels
def read_images_from_disk(input_queue):
"""Consumes a single filename and label as a ' '-delimited string.
Parameters
----------
filename_and_label_tensor: A scalar string tensor.
Returns
-------
Two tensors: the decoded image, and the string label.
"""
label = input_queue[1]
file_contents = tf.read_file(input_queue[0])
example = tf.image.decode_png(file_contents, channels=3)
# example = rescale_image(example)
# processed_label = label
return example, label
def random_resize(image, lower_size, upper_size):
"""Randomly resizes an image
Parameters
----------
lower_size:
upper_size:
Returns
-------
a randomly resized image
"""
new_size = tf.to_int32(
tf.random_uniform([], lower_size, upper_size))
return tf.image.resize_images(image, new_size, new_size,
method=0)
def _input_pipeline(filename, batch_size,
processing_image=lambda x: x,
processing_label=lambda y: y,
num_epochs=None):
"""The input pipeline for reading images classification data.
The data should be stored in a single text file of using the format:
/path/to/image_0 label_0
/path/to/image_1 label_1
/path/to/image_2 label_2
...
Args:
filename: the path to the txt file
batch_size: size of batches produced
num_epochs: optionally limited the amount of epochs
Returns:
List with all filenames in file image_list_file
"""
# Reads pfathes of images together with there labels
image_list, label_list = read_labeled_image_list(filename)
images = tf.convert_to_tensor(image_list, dtype=tf.string)
labels = tf.convert_to_tensor(label_list, dtype=tf.int32)
# Makes an input queue
input_queue = tf.train.slice_input_producer([images, labels],
num_epochs=num_epochs,
shuffle=True)
# Reads the actual images from
image, label = read_images_from_disk(input_queue)
pr_image = processing_image(image)
pr_label = processing_label(label)
image_batch, label_batch = tf.train.batch([pr_image, pr_label],
batch_size=batch_size,
shapes = [256,256,3])
# Display the training images in the visualizer.
tensor_name = image.op.name
tf.image_summary(tensor_name + 'images', image_batch)
return image_batch, label_batch
def test_pipeline():
data_folder = '/home/kang/Documents/work_code_PC1/data/UCLandUsedImages/'
data_file = 'UCImage_Labels.txt'
filename = os.path.join(data_folder, data_file)
image_batch, label_batch = _input_pipeline(filename, 75)
# Create the graph, etc.
init_op = tf.initialize_all_variables()
sess = tf.InteractiveSession()
sess.run(init_op)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
a = sess.run([image_batch, label_batch])
coord.request_stop()
coord.join(threads)
print("Finish Test")
return a
if __name__ == '__main__':
# aa = test_preprocc()
# matplotlib.pyplot.imshow(aa[1])
a1 = test_pipeline()
a2 = test_pipeline()
但是它出错了，它让我困惑了很长时间：
Traceback (most recent call last):
File "<ipython-input-7-e24901ce3365>", line 1, in <module>
runfile('/home/kang/Documents/work_code_PC1/VGG_tensorflow_UCMerced/readUClandUsedImagetxt1.py', wdir='/home/kang/Documents/work_code_PC1/VGG_tensorflow_UCMerced')
File "/usr/local/lib/python2.7/dist-packages/spyderlib/widgets/externalshell/sitecustomize.py", line 714, in runfile
execfile(filename, namespace)
File "/usr/local/lib/python2.7/dist-packages/spyderlib/widgets/externalshell/sitecustomize.py", line 81, in execfile
builtins.execfile(filename, *where)
File "/home/kang/Documents/work_code_PC1/VGG_tensorflow_UCMerced/readUClandUsedImagetxt1.py", line 254, in <module>
a1 = test_pipeline()
File "/home/kang/Documents/work_code_PC1/VGG_tensorflow_UCMerced/readUClandUsedImagetxt1.py", line 244, in test_pipeline
a = sess.run([image_batch, label_batch])
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 340, in run
run_metadata_ptr)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 564, in _run
feed_dict_string, options, run_metadata)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 637, in _do_run
target_list, options, run_metadata)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 659, in _do_call
e.code)
InvalidArgumentError: Different number of component types. Types: uint8, int32, Shapes: [[256,256,3]]
[[Node: batch_11/fifo_queue = FIFOQueue[capacity=32, component_types=[DT_UINT8, DT_INT32], container="", shapes=[[256,256,3]], shared_name="", _device="/job:localhost/replica:0/task:0/cpu:0"]()]]
Caused by op u'batch_11/fifo_queue', defined at:
答案 0 :(得分：0)
错误是由函数tf.train.batch的错误参数shapes引起的。参数shapes应该保留为默认值，或者应该是：
shapes :(可选）每个示例的形状。默认为tensor_list
的推断形状
在这里您要shapes = [256, 256, 3]，但是您应该在列表中指定pr_image和pr_label的形状：
image_batch, label_batch = tf.train.batch(
[pr_image, pr_label],
batch_size=batch_size,
shapes = [[256,256,3], pr_label.get_shape()])