The difference bewteen tf.shape() and tensor.get_shape()

https://stackoverflow.com/questions/36966316/how-to-get-the-dimensions-of-a-tensor-in-tensorflow-at-graph-construction-time

I see most people confused about tf.shape(tensor) and tensor.get_shape() Let's make it clear:

1. tf.shape

tf.shape is used for dynamic shape. If your tensor's shape is changable, use it. An example: a input is an image with changable width and height, we want resize it to half of its size, then we can write something like:
new_height = tf.shape(image)[0] / 2

2. tensor.get_shape

tensor.get_shape is used for fixed shapes, which means the tensor's shape can be deduced in the graph.

Conclusion: tf.shape can be used almost anywhere, but t.get_shape only for shapes can be deduced from graph.

def shape(tensor):
    s = tensor.get_shape()
    return tuple([s[i].value for i in range(0, len(s))])

Example:

batch_size, num_feats = shape(logits)






















https://stackoverflow.com/questions/43563609/how-tf-transpose-works-in-tensorflow




Looking at the numpy.transpose documentation, we find that transpose takes the argument

axes : list of ints, optional
By default, reverse the dimensions, otherwise permute the axes according to the values given.

So the default call to transpose translates into np.transpose(a, axes=[1,0]) for the 2D case, or np.transpose(a, axes=[2,1,0]).

The operation you want to have here, is one that leaves the "depth" dimension unchanged. Therefore in the axes argument, the depth axes, which is the 0th axes, needs to stay unchanged. The axes 1 and 2 (where 1 is the vertical axis), need to change positions. So you change the axes order from the initial [0,1,2] to [0,2,1] ([stays the same, changes with other, changes with other]).

In tensorflow, they have for some reason renamed axes to perm. The argument from above stays the same.

images

Concerning images, they differ from the arrays in the question. Images normally have their x and y stored in the first two dimensions and the channel in the last, [y,x,channel].

In order to "transpose" an image in the sense of a 2D transposition, where horizontal and vertical axes are exchanged, you would need to use

np.transpose(a, axes=[1,0,2])

(channel stays the same, x and y are exchanged).

https://stackoverflow.com/questions/36966316/how-to-get-the-dimensions-of-a-tensor-in-tensorflow-at-graph-construction-time

Let's make it simple as hell. If you want a single number for the number of dimensions like 2, 3, 4, etc., then just use tf.rank(). But, if you want the exact shape of the tensor then use tensor.get_shape()

with tf.Session() as sess:
   arr = tf.random_normal(shape=(10, 32, 32, 128))
   a = tf.random_gamma(shape=(3, 3, 1), alpha=0.1)
   print(sess.run([tf.rank(arr), tf.rank(a)]))
   print(arr.get_shape(), ", ", a.get_shape())     


# for tf.rank()    
[4, 3]

# for tf.get_shape()
Output: (10, 32, 32, 128) , (3, 3, 1)