AttributeError: 'FrameIndexSplitter' object has no attribute '_id'

[zamanmub]

Hello,
I am trying object detection example, but getting the following error:

File "C:\Program Files\Python37\lib\multiprocessing\spawn.py", line 115, in _main self = reduction.pickle.load(from_parent) File "C:\Users\Mubtaseem Zaman\AppData\Roaming\Python\Python37\site-packages\videoflow\core\node.py", line 40, in __hash__ return self._id AttributeError: 'FrameIndexSplitter' object has no attribute '_id'

[jadielam]

@mubtaseemz What version of the code are you using?

[zamanmub]

Thanks for prompt reply. I am using 0.2.2 video flow. Python 3.7. Tensorflow 1.13

[flappy007]

change self._id to id(self) in the hash function in node.py

[jadielam]

@flappy007 I have not been able to reproduce this issue.

The decision to implement the hash function in that way is the following: a node object will be accessed by different operating system processes. I don't think that there is a guarantee that the id function will return the same value across processes (but I might be wrong). Because of that I am calling the id() function at object construction time.

[tunght]

I have the same issue, i'm using Mac and Python3.8
Change self._id to id(self) in the hash function does do the trick, but it also does generate a different value in the new process