Tqdm error (on testing) --> Batch size > 1

[Utena]

Dear Rodney LaLonde,

First of all, I would like to thank you for this amazing work and for publishing your source code! :) I really appreciate your contribution. I am also working on the field of "Capsule networks" for segmentation. I obtained some satisfactory results running your Capsr3 network with a batch size = 1, but when I increase this parameter (i.e., batch size = 2) the code crashes! More specifically, the following error appears only on testing time:

38/75 [==============>...............] - ETA: 5sException KeyError: KeyError(<weakref at 0x7f88e74e5f70; to 'tqdm' at 0x7f88e7500d90>,) in <bound method tqdm.__del__ of   0%|                                                                          | 0/18 [00:05<?, ?it/s]> ignored
Traceback (most recent call last):
  File "main.py", line 169, in <module>
    main(arguments)
  File "main.py", line 81, in main
    test(args, test_list, model_list, net_input_shape)
  File "/home/.../.../SegCapsOfficial/test.py", line 139, in test
    use_multiprocessing=False, verbose=1)
  File "/home/.../.../tf9.1/lib/python2.7/site-packages/keras/legacy/interfaces.py", line 87, in wrapper
    return func(*args, **kwargs)
  File "/home/.../.../tf9.1/lib/python2.7/site-packages/keras/engine/training.py", line 2407, in predict_generator
    generator_output = next(output_generator)
  File "/home/.../.../tf9.1/lib/python2.7/site-packages/keras/utils/data_utils.py", line 727, in get
    raise StopIteration()
StopIteration

I tried to solve it following this post: , but the error still persists and it is not possible to run the networks with a batch size > 1! Is someone also concerned about this bug? My Python and Tqdm versions are these ones:

Python 2.7.14
Tqdm 4.23.3

Thank you folks! ;)
Jordina

[lalonderodney]

Hello @Utena ,

I have not encountered this issue. My images are so large I can only run with batch size of 1 but I have other versions of this code which run with batch size > 1 no problem (for other projects I am working on ;) ). Is anyone else able to reproduce this bug?

[Utena]

Good morning @lalonderodney !

Thank you very much for your response :) Do you have the same Tqdm 4.23.3 version like me? It is a little bit strange, since this error only appears in testing time... >.< I can train my capsule network with no bugs (images of size 224 x 224).

Cheers :-* !
Jordina