Knowledge Distillation on backbone models

Hi,

Ideally, you should be able to train a student model with the implementation like below:

And the output model will have a similar architecture as the teacher SSD MobileNet.
So you can deploy it on Jetson.

Thanks.