How to use numpy arrays as input to Tensorflow CNN without mismatched dimensions

Question

I have inputs to a tensorflow convnet as rank-4 tensors (32, 32, 3, 73257) (73257 comes from number of imgs) which are numpy arrays, but my placeholder variable for my x inputs is 2-d, (None, 3072). The 3072 comes from the img height x img width x channels. My question is, how do I reshape or use the images so that theyre compatible with the placeholder?

P.S. These images are from the SVHN cropped 32x32 dataset

images = np.array(features, dtype=np.float32)
...
x = tf.placeholder(tf.float32, shape=[None, 3072])
...
for _ in range(1000):
  batch = next_batch(50, images, labels)
  train_step.run(feed_dict={x: batch[0], y_: batch[1]})
...
with tf.Session() as sess:
  sess.run(tf.global_variables_initializer())
  for i in range(20000):
    batch = next_batch(50, images, labels)
    if i % 100 == 0:
      train_accuracy = accuracy.eval(feed_dict={
          x: batch[0], y_: batch[1], keep_prob: 1.0})
      print('step %d, training accuracy %g' % (i, train_accuracy))
    train_step.run(feed_dict={x: images, y_: labels, keep_prob: 0.5})

pgross · Accepted Answer · 2017-09-15 04:28:12Z

3

Assuming you have 73257 images 32 by 32 pixel holding 3 bands (e.g. rgb). You can do a

input = tf.transpose(input, [3, 0, 1, 2])

to bring the last dimension first. The tensor should then look like (73257, 32, 32, 3).

Then do

input = tf.reshape(input, [-1, 3072])

to reduce the dimension. The tensor should then look like (73257, 3072).

answered Sep 15, 2017 at 4:28

pgross

5095 silver badges16 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Meka Over a year ago

Got it! Thanks for your help

Collectives™ on Stack Overflow

How to use numpy arrays as input to Tensorflow CNN without mismatched dimensions

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related