我想使用tensor-flow打印我自己图像的所有概率。以下代码适用于我从在线获得的任何mnist图像,但不适用于我自己的图像。这是整个代码。
from PIL import Image
import random
from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf
import numpy as np
import scipy.ndimage
from PIL import Image
np.set_printoptions(threshold='nan')
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)
x = tf.placeholder(tf.float32, [None, 784])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
y = tf.nn.softmax(tf.matmul(x, W) + b)
y_ = tf.placeholder(tf.float32, [None, 10])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)
init = tf.initialize_all_variables()
sess = tf.Session()
sess.run(init)
for i in range(1000):
batch_xs, batch_ys = mnist.train.next_batch(1000)
sess.run(train_step, feed_dict= {x: batch_xs, y_: batch_ys})
print ("done with training")
print ("\n\n\n")
probabilities = y
img = Image.open('mnist_7_367.jpg').convert('L') # convert image to 8-bit grayscale
data = list(img.getdata()) # convert image data to a list of integers
newList = [float(x) / 255.0 for x in data]
newList1 = np.asarray(newList)
batch_xs = np.reshape(newList1, (1, 784))
batch_xs[batch_xs < 0.1] = 0
#print isinstance(newList1, list)
print batch_xs
counter = 0
for x in probabilities.eval(feed_dict={x: batch_xs}, session = sess):
print ('[')
for y in x:
print (counter, '{:f}'.format(float(y)))
counter += 1
print (']')
hello = np.reshape(batch_xs, (28, 28))
Matrix = np.array([])
for x in hello:
for y in x:
Matrix = np.append(Matrix, int(255 * y))
imageReal = np.reshape(Matrix, (28, 28))
img = Image.fromarray(imageReal)
img.show()
sess.close()
我遇到的问题是
for x in probabilities.eval(feed_dict={x: batch_xs}, session = sess):
产生以下错误:
Cannot interpret feed_dict key as Tensor: Can not convert a int into a Tensor.
我觉得问题出在session = sess部分行。