Open hnwynters opened 2 years ago
This cannot be done the way darknet is set up right now.
The best way to create a model that knows all three classes is by training a model with all three classes to begin with. There is no better solution/real alternative for object detection with darknet.
I would like to train, for example, three classes; turtles, koalas, and cats. But for several reasons I would prefer to train only one class at a time. I would like to end up with a training file that understands and can detect all three classes. I understand how to train one class at a time and I understand how to train all three classes at a time. But how to add training to an existing training file I do not understand.
Can someone please explain how this is done?
Thank you.