titu1994/neural-image-assessment

memory leak when evaluating

Closed this issue · 4 comments

swig/python detected a memory leak of type 'int64_t *', no destructor found.

When does this occur ? This is not much info to go on.

using Python2 instead of Python3 to eval the pre-trained model, such as mobileNet. In the script, I attach the code "from future import division,print_function" , and copy a path.py from python3's lib. When evaluating the images (-img for single image or -dir for picture set), the script performs well and does give the score, but the error (swig/python detected a memory leak of type 'int64_t *', no destructor found.) shows up in the last line.

This is an error from python internals, and may most probably a combination of using python 2 with a pythin 3.5 library. I won't be able to fix such an issue.

I recommend upgrading to pythin 3.5 in the near future, numpy will be dropping support for Python 2 I'm a few years.

i just update my python to version3.6, but i still get this "memory leak" error. Just let you know. maybe it's my machine env problem...