FYI, if you run this code on windows, you are effectively iterating 2X times.
The code example was based on linux which has case sensitive filenames. Since windows does not, the create_image_lists() function returns a list containing the filenames with JPEG and jpeg.
Change line 611 of retrain.py from extensions = ['jpg', 'jpeg', 'JPG', 'JPEG'] to extensions = ['jpg', 'jpeg'] to avoid training twice.
FYI, if you run this code on windows, you are effectively iterating 2X times.
The code example was based on linux which has case sensitive filenames. Since windows does not, the create_image_lists() function returns a list containing the filenames with JPEG and jpeg.
Change line 611 of retrain.py from
extensions = ['jpg', 'jpeg', 'JPG', 'JPEG']toextensions = ['jpg', 'jpeg']to avoid training twice.