ValueError: Cannot feed value of shape (0,) for Tensor 'Placeholder_1:0', which has shape '(?, 180)'

[861]

Hi, I use tensorflow（1.2.1），it‘s my first time to using tensorflow，I just want to run your code for learing，but I met the error：

Traceback (most recent call last):
File "E:\captcha-breaker-master\captcha_train.py", line 252, in
train()
File "E:\captcha-breaker-master\captcha_train.py", line 218, in train
smry, _, l = sess.run([merged, optimizer, loss], feed_dict={x: batch[0], y: batch_labels})
File "C:\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 789, in run
run_metadata_ptr)
File "C:\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 975, in _run
% (np_val.shape, subfeed_t.name, str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (0,) for Tensor 'Placeholder_1:0', which has shape '(?, 180)'

[vitorarins]

Hi @861. Do you have images to train the model? Can you send me a sample of those images?

You will need about 500k images properly labeled. If you want I can help you in the process of training the model.

[861]

Yes，I have labeld about 500k images in train folder，you can give me the email ，so I can send them to you！

[ghost]

I'm using TensorFlow 1.4.0 and wanted to use your code to learn and eventually try to play around with.
Oddly enough, your code (with some name changes etc.) did work for me once, but my CPU was overstrained, so I decided to run it with my GPU (CUDA 8.0) - since then, I get the same error message on both CPU and GPU.

[ghost]

After playing around a bit I was able to solve this problem by using the absolute paths to the train and test folders (as well as the summary directory just for safety).
Sadly I am not very good with programming in Python, so I cannot say why your code does not work in this regard. I also do not know if my solution makes it slower etc. but at least this problem is solved :D