Tensorflow Slim：TypeError：预期 int32，得到的列表包含类型为“_Message”的张量

Question

我正在关注 this 学习 TensorFlow Slim 的教程，但在运行以下 Inception 代码时：

import numpy as np
import os
import tensorflow as tf
import urllib2

from datasets import imagenet
from nets import inception
from preprocessing import inception_preprocessing

slim = tf.contrib.slim

batch_size = 3
image_size = inception.inception_v1.default_image_size
checkpoints_dir = '/tmp/checkpoints/'
with tf.Graph().as_default():
    url = 'https://upload.wikimedia.org/wikipedia/commons/7/70/EnglishCockerSpaniel_simon.jpg'
    image_string = urllib2.urlopen(url).read()
    image = tf.image.decode_jpeg(image_string, channels=3)
    processed_image = inception_preprocessing.preprocess_image(image, image_size, image_size, is_training=False)
    processed_images  = tf.expand_dims(processed_image, 0)

    # Create the model, use the default arg scope to configure the batch norm parameters.
    with slim.arg_scope(inception.inception_v1_arg_scope()):
        logits, _ = inception.inception_v1(processed_images, num_classes=1001, is_training=False)
    probabilities = tf.nn.softmax(logits)

    init_fn = slim.assign_from_checkpoint_fn(
        os.path.join(checkpoints_dir, 'inception_v1.ckpt'),
        slim.get_model_variables('InceptionV1'))

    with tf.Session() as sess:
        init_fn(sess)
        np_image, probabilities = sess.run([image, probabilities])
        probabilities = probabilities[0, 0:]
        sorted_inds = [i[0] for i in sorted(enumerate(-probabilities), key=lambda x:x[1])]

    plt.figure()
    plt.imshow(np_image.astype(np.uint8))
    plt.axis('off')
    plt.show()

    names = imagenet.create_readable_names_for_imagenet_labels()
    for i in range(5):
        index = sorted_inds[i]
        print('Probability %0.2f%% => [%s]' % (probabilities[index], names[index]))

我似乎遇到了这组错误：

Traceback (most recent call last):
  File "DA_test_pred.py", line 24, in <module>
    logits, _ = inception.inception_v1(processed_images, num_classes=1001, is_training=False)
  File "/home/deepankar1994/Desktop/MTP/TensorFlowEx/TFSlim/models/slim/nets/inception_v1.py", line 290, in inception_v1
    net, end_points = inception_v1_base(inputs, scope=scope)
  File "/home/deepankar1994/Desktop/MTP/TensorFlowEx/TFSlim/models/slim/nets/inception_v1.py", line 96, in inception_v1_base
    net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/array_ops.py", line 1053, in concat
    dtype=dtypes.int32).get_shape(
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/ops.py", line 651, in convert_to_tensor
    as_ref=False)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/ops.py", line 716, in internal_convert_to_tensor
    ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/constant_op.py", line 176, in _constant_tensor_conversion_function
    return constant(v, dtype=dtype, name=name)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/constant_op.py", line 165, in constant
    tensor_util.make_tensor_proto(value, dtype=dtype, shape=shape, verify_shape=verify_shape))
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/tensor_util.py", line 367, in make_tensor_proto
    _AssertCompatible(values, dtype)
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/tensor_util.py", line 302, in _AssertCompatible
    (dtype.name, repr(mismatch), type(mismatch).__name__))
TypeError: Expected int32, got list containing Tensors of type '_Message' instead.

这很奇怪，因为所有这些代码都来自他们的官方指南。我是 TF 的新手，如有任何帮助，我们将不胜感激。

Answer 1

我发现大多数人的回答都是错误的。这只是由于 tf.concat 的变化。 它以下列方式工作。

net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])

使用下面的

net = tf.concat(values=[branch_0, branch_1, branch_2, branch_3],axis=3,)

记住，传递关键字参数时应该在其他参数之前。

Answer 2

我在使用 1.0 版本时遇到了同样的问题，我可以让它工作而不必回滚到以前的版本。

这个问题是由于 api 的变化引起的。那次讨论帮助我找到了解决方案：Google group > Recent API Changes in TensorFlow

你只需要用 tf.concat 更新所有的行

例如

net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])

应该改为

net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)

注意：

我能够毫无问题地使用这些模型。但是之后我想加载预训练的重量时仍然出错。 自从他们制作了检查点文件以来，似乎 slim 模块发生了一些变化。代码创建的图形与检查点文件中存在的图形不同。

注2：

我能够通过添加到所有 conv2d 层 biases_initializer=None 来使用 inception_resnet_v2 的预训练权重

Answer 3

明确写出参数的名称可以解决问题。

而不是

net = tf.concat(3, [branch_0, branch_1, branch_2, branch_3])

使用

net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])

Answer 4

我在做这项工作时遇到了同样的错误。

我发现了

logits = tf.nn.xw_plus_b(tf.concat(outputs, 0), w, b)
loss = tf.reduce_mean(
  tf.nn.softmax_cross_entropy_with_logits(
    labels=tf.concat(train_labels, 0), logits=logits))

输出是shape=(10, 64, 64)。

代码希望将输出[0] 连接到输出[9] => 获得一个新形状(640,64)。

但“tf.concat”API 可能不允许这样做。

(train_labels 与此相同)

所以我写信给

A = tf.concat(0,[outputs[0],outputs[1]])
A = tf.concat(0,[A,outputs[2]])
A = tf.concat(0,[A,outputs[3]])
A = tf.concat(0,[A,outputs[4]])
A = tf.concat(0,[A,outputs[5]])
A = tf.concat(0,[A,outputs[6]])
A = tf.concat(0,[A,outputs[7]])
A = tf.concat(0,[A,outputs[8]])
A = tf.concat(0,[A,outputs[9]])
B = tf.concat(0,[train_labels[0],train_labels[1]])
B = tf.concat(0,[B,train_labels[2]])
B = tf.concat(0,[B,train_labels[3]])
B = tf.concat(0,[B,train_labels[4]])
B = tf.concat(0,[B,train_labels[5]])
B = tf.concat(0,[B,train_labels[6]])
B = tf.concat(0,[B,train_labels[7]])
B = tf.concat(0,[B,train_labels[8]])
B = tf.concat(0,[B,train_labels[9]])

logits = tf.nn.xw_plus_b(tf.concat(0, A), w, b)
loss = tf.reduce_mean(
  tf.nn.softmax_cross_entropy_with_logits(
    labels=tf.concat(0, B), logits=logits))

它可以运行！