ichibod / object-detection-for-coreml Goto Github PK

Hello!

I'm hoping you can help me here...
I've gotten down to the part where we are prepping and training, however, I'm noticing that after running prep.py, the Turi Create visualization tool has an empty dictionary of annotations for each image.

I know this is probably user error, but is there anything you might suspect I'm doing here to cause this?
Thanks,
David

I receive the following error when running prep.py:

Traceback (most recent call last):
  File "prep.py", line 33, in <module>
    data['annotations'] = SArray(data=annotations, dtype=list)
  File "/home/rpinglet/venv/local/lib/python2.7/site-packages/turicreate/data_structures/sframe.py", line 3585, in __setitem__
    self.add_column(sa_value, key, inplace=True)
  File "/home/rpinglet/venv/local/lib/python2.7/site-packages/turicreate/cython/context.py", line 49, in __exit__
    raise exc_type(exc_value)
RuntimeError: Column "annotations" has different size than current columns!

I used the Simple Image Annotator, which worked well. The convert.py also seems to have worked as expected. Is there a change that needs to be made to annotations after convert.py has completed?