ValueError: Shape must be rank 2 but is rank 3 for 'MatMul' (op: 'MatMul') with input shapes: [?,1000,2048], [2048,64].

[zhangxgu]

The error raised when I run this code:
'''
import loupe as lp
import tensorflow as tf
x = tf.placeholder("float", [None,1000,2048])
NetVLAD = lp.NetVLAD(feature_size=2048, max_samples=1000, cluster_size=64,
output_dim=2048, gating=True, add_batch_norm=True,
is_training=True)
NetVLAD.forward(x)
'''
I think this x.shape is #batch_size dot #max_sample dot #feature_size. Should I change line 126 in loupe.py into
'''
cluster_weights = tf.get_variable("cluster_weights",
[1, self.feature_size, self.cluster_size],
initializer = tf.random_normal_initializer(
stddev=1 / math.sqrt(self.feature_size)))
''' ?
But it also lead to other error, can you help me?
Thank you very much!

[antoine77340]

Hi,
You are right sorry there was an error in my documentation. I am changing that.
So in the documentation I wrote:
Args:
reshaped_input: The input in reshaped in the following form:
'batch_size' x 'max_samples' x 'feature_size'.

But I meant:

   Args:
    reshaped_input: Your input of form:
    'batch_size' x 'max_samples' x 'feature_size'.
    must be reshaped in that form:
     'batch_size*max_samples' x 'feature_size' by performing:
     reshaped_input = tf.reshape(input, [-1, feature_size])

So if you instead do
x = tf.placeholder("float", [1,1000,2048]) # you can put any batch size and not only 1
x = tf.reshape(x, [-1,2048])
NetVLAD = lp.NetVLAD(feature_size=2048, max_samples=1000, cluster_size=64,
output_dim=2048, gating=True, add_batch_norm=True,
is_training=True)
NetVLAD.forward(x)

It should work.

Thank you for pointing this.

Antoine