I want to know whether the complete YOLOv8n training process can be performed on the OAK4/RVC4 itself, without using an external PC or GPU.
If direct training is not supported, what is the recommended workflow for training YOLOv8n and then deploying the trained model for inference on the OAK4?