Cannot feed value of shape () for Tensor 'input/x-input:0', which ha s shape '(1, 3072)' in TensorFlow

[hitesha14]

Hello Joel Lutman,@jhole89

I get the following shape error when I am trying to run the CNN Classifier code. However I have tried all means of resolving it, but I wasn't able to resolve it at all. I have few images that I am trying to classify. I am using image size as 32. I have tried with different image sizes on my custom dataset, however I face similar issue everytime.

Please see the below stack trace of the error. I would be very grateful if you could please help me out with it.

INFO:root:Loading resource: Images [C:\Users\hits\cnn_classifier\Output\images\t
rain]
img_size 3072
length of data 32
WARNING:tensorflow:From C:\Users\hits\cnn_classifier\cnn_model.py:171: softmax_c
ross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and w
ill be removed in a future version.
Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow
into the labels input on backprop by default.

See @{tf.nn.softmax_cross_entropy_with_logits_v2}.

WARNING:tensorflow:From C:\Users\hits\cnn_classifier\cnn_model.py:171: softmax_c
ross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and w
ill be removed in a future version.
Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow
into the labels input on backprop by default.

See @{tf.nn.softmax_cross_entropy_with_logits_v2}.

WARNING:tensorflow:From C:\Users\hits\cnn_classifier\cnn_model.py:189: calling a
rgmax (from tensorflow.python.ops.math_ops) with dimension is deprecated and wil
l be removed in a future version.
Instructions for updating:
Use the axis argument instead
WARNING:tensorflow:From C:\Users\hits\cnn_classifier\cnn_model.py:189: calling a
rgmax (from tensorflow.python.ops.math_ops) with dimension is deprecated and wil
l be removed in a future version.
Instructions for updating:
Use the axis argument instead
2018-07-20 16:08:52.728842: I T:\src\github\tensorflow\tensorflow\core\platform
cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow bi
nary was not compiled to use: AVX2
WARNING:root:Resource not found: CNN Model [C:\Users\hits\cnn_classifier\Output
tmp\tensorflow/cnn/model\model.ckpt]. Model will now be trained from scratch.
Tensor("cost/Mean:0", shape=(), dtype=float32) {<tf.Tensor 'input/x-input:0' sha
pe=(1, 3072) dtype=float32>: 1, <tf.Tensor 'input/y_true:0' shape=(1, 2) dtype=f
loat32>: 1, <tf.Tensor 'input/Placeholder:0' shape= dtype=float32>: 1.0
}
Traceback (most recent call last):
File "main.py", line 47, in
train(image_directory, model_directory)
File "C:\Users\hits\cnn_classifier\cnn_model.py", line 264, in train
checkpoint_path=os.path.join(checkpoint_dir, 'model.ckpt'))
File "C:\Users\hits\cnn_classifier\cnn_model.py", line 79, in log_progress
val_loss = session.run(cost, feed_dict=test_feed_dict)
File "C:\Users\hits\Anaconda3\lib\site-packages\tensorflow\python\client\sessi
on.py", line 900, in run
run_metadata_ptr)
File "C:\Users\hits\Anaconda3\lib\site-packages\tensorflow\python\client\sessi
on.py", line 1111, in _run
str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape () for Tensor 'input/x-input:0', which ha
s shape '(1, 3072)'

@jhole89