Tensorflow custom object detector numpy error

Question

I've been working through the tutorial at https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/training.html trying to make an object detector. I've been through all the steps I believe correctly, but when I run the train commmand...

python model_main_tf2.py --model_dir=models/my_ssd_resnet50_v1_fpn --pipeline_config_path=models/my_ssd_resnet50_v1_fpn/pipeline.config

...in my conda prompt this is the error. I'm including the end of the traceback. I can give more if need be.

... File "pycocotools_mask.pyx", line 1, in init pycocotools._mask ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

I don't know what binary incompatibility is, and every other SO post I've seen regarding this error wants it to be a mismatch somewhere between versions of libraries I've pipped. That may still be what it is, but I've gone through a bunch of different version combinations trying to get it going and no luck. Any help is appreciated.

Elkelthen · Accepted Answer · 2021-02-19 13:46:52Z

0

Figured out this error, anyway. I reinstalled all of the pip libraries I had installed.

pip freeze --local >pipfreeze.txt
pip install --upgrade --force-reinstall -r pipfreeze.txt

That seems to have fixed it. I have other errors later in the process now, so this may not be the final solution for someone else, but that's that.

answered Feb 19, 2021 at 13:46

Elkelthen

1711 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Tensorflow custom object detector numpy error

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related