Community
cancel
Showing results for 
Search instead for 
Did you mean: 
idata
Community Manager
276 Views

NCS1: Fine-tune TinyYOLO model for custom dataset

I was wondering if there was an easy way to fine-tune the TinyYOLO model to detect objects from a custom dataset in real time with bounding boxes and then compile the Caffe files into the NCS graph file. There are 4 classes in my dataset with about 150 images for each class so it is not a big dataset.

0 Kudos
3 Replies
idata
Community Manager
99 Views

@user123

 

I trained with a custom dataset based on the following repository.

 

https://github.com/duangenquan/YoloV2NCS.git
idata
Community Manager
99 Views

What file in the repository do I need to edit to have it train on my own dataset?

idata
Community Manager
99 Views

And I should probably place bounding boxes on the objects in my images? If so, what did you use to do that? Thanks

Reply