代码之家 › 专栏 › 技术社区 › user5593270

Tensorflow值错误:要解包的值太多(预期为2)

mnist generator runtime-error tensorflow python

2

user5593270 · 技术社区 · 9 年前

作为参考,我使用 Python 3 TensorFlow

我正在尝试使用我自己的数据集(300张猫图片,512x512.png格式) Tensorflow

ValueError: too many values to unpack (expected 2) 。错误出现在行中 images,labal = create_batches(10) ,它指向我的函数 create_batches TensorFlow .我正在尝试基于MNIST数据集创建自己的神经网络。代码如下:

import tensorflow as tf
import numpy as np
import os
import sys
import cv2


content = []
labels_list = []
with open("data/cats/files.txt") as ff:
    for line in ff:
        line = line.rstrip()
        content.append(line)

with open("data/cats/labels.txt") as fff:
    for linee in fff:
        linee = linee.rstrip()
        labels_list.append(linee)

def create_batches(batch_size):
    images = []
    for img in content:
        #f = open(img,'rb')
        #thedata = f.read().decode('utf8')
        thedata = cv2.imread(img)
        thedata = tf.contrib.layers.flatten(thedata)
        images.append(thedata)
    images = np.asarray(images)

    labels =tf.convert_to_tensor(labels_list,dtype=tf.string)

    print(content)
    #print(labels_list)

    while(True):
        for i in range(0,298,10):
            yield images[i:i+batch_size],labels_list[i:i+batch_size]


imgs = tf.placeholder(dtype=tf.float32,shape=[None,262144])
lbls = tf.placeholder(dtype=tf.float32,shape=[None,10])

W = tf.Variable(tf.zeros([262144,10]))
b = tf.Variable(tf.zeros([10]))

y_ = tf.nn.softmax(tf.matmul(imgs,W) + b)

cross_entropy = tf.reduce_mean(-tf.reduce_sum(lbls * tf.log(y_),reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.05).minimize(cross_entropy)

sess = tf.InteractiveSession()
tf.global_variables_initializer().run()
for i in range(10000):
    images,labal = create_batches(10)
    sess.run(train_step, feed_dict={imgs:images, lbls: labal})

correct_prediction = tf.equal(tf.argmax(y_,1),tf.argmax(lbls,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))

print(sess.run(accuracy, feed_dict={imgs:content, lbls:labels_list}))

以及错误:

Traceback (most recent call last):
  File "B:\Josh\Programming\Python\imgpredict\predict.py", line 54, in <module>

    images,labal = create_batches(2)
ValueError: too many values to unpack (expected 2)
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile
(A few hundred lines of this)
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile

GitHub link 如果有人需要,请链接。项目文件夹是“imgpredict”。

2 回复 | 直到 9 年前

1

2

DarkCygnus 9 年前

yield(images[i:i+batch_size]) #,labels_list[i:i+batch_size])

这给了您一个生成的值,但当您调用方法时,您期望生成两个值:

images,labal = create_batches(10)

其中一个产生两个值,例如:

yield (images[i:i+batch_size] , labels_list[i:i+batch_size])

(取消注释)或者只需要一个。

编辑

#when yielding, remember that yield returns a Generator, therefore the ()
yield (images[i:i+batch_size] , labels_list[i:i+batch_size])

#When receiving also, even though this is not correct
(images,labal) = create_batches(10)

然而这不是我使用 yield 结束返回生成器的方法,在您的示例中应该如下所示:

#do the training several times as you have
for i in range(10000):
    #now here you should iterate over your generator, in order to gain its benefits
    #that is you dont load the entire result set into memory
    #remember to receive with () as mentioned
    for (images, labal) in create_batches(10):
        #do whatever you want with that data
        sess.run(train_step, feed_dict={imgs:images, lbls: labal})

this 关于用户的问题 产量

2

0

Prune 9 年前

您注释掉了第二个退货项目。

        yield(images[i:i+batch_size])    #,labels_list[i:i+batch_size])

images labal 。删除该注释标记,如果处于调试模式,则生成一个伪值。

更新

result = (images[i:i+batch_size],
          labels_list[i:i+batch_size])
print len(result), result
return result