Dene
Thanks for clarifying all the stuff. Reload add-on with shortcut is kinda alright, but not the best way, because one needs to push button every time add-on is changed. So...
I have the same error.
Provide more information please. Step, video, etc.
I've checked and everything seems fine. Check Runtime > Change runtime type. That should be Python 3 with GPU.
Video upload is finished, right? Do you run next cell after the video is uploaded completely?
I think the error may appear because of the wrong tensorflow version. `1.3.0` is needed. I'll look into this.
There are two versions of Python used here, that's not the problem I guess. Please, try changing `config.py` and let me know about the results. Thanks.
> Today I tested the change I mentioned above, replacing the three occurrences of 'log_dir' with 'logging_dir' in config.py and that fixed the problem with duplicate definitions. In order to...
Out of the box - no. You could use something like AlphaPose for tracking first - extracting bounding boxes and crop these parts from video. After that you could provide...
Provide more information, please.