Question: finetune 13 class with custom dataset

[jerryho-quanta]

Hi Eric,

Thanks your git,

I want to finetune 13 class with custom dataset with darknet 53 yolov3 from https://github.com/eric612/Caffe-YOLOv3-Windows After changing the related layer( link ) about class, do I need also freeze the darnet 53 backbone part of yolov3? It seems that your train.prototxt doesn't freeze the darnet 53 backbone part?

Thanks,

Jerry Ho