TX2 Training New Model with Digit (ParseError)

Autonomous Machines Jetson & Embedded Systems Jetson TX2
1

Hi all,

I was following the tutorial at jetson-inference to train a detectnet using MS-COCO dataset but I encountered an issue where I got this error box popped up on Digit "ParseError 72:3 : Message type “caffe.LayerParameter” has no field named “detectnet_groundtruth_param”.

The training is done on TX2 board itself as I currently dont have a proper host computer to run Digits. I understand it is advised to do the training on a host computer and transfer the weight to Jetson for inference.

I will appreciate it if someone can shed some light to this issue.

Thanks in advance.

2

Hi,

May I know which Caffe branch do you use?
Please remember that DIGITs requires nvcaffe-0.15.

Thanks.

3

Hi,

I am assuming it is the latest one. As I git clone it from the repo.
Also just to re-clarify I am trying to run DIGITs on TX2.

Thanks.

4

Hi,

DIGITs doesn’t support Jetson platform due to the lack of nvidia-smi.
Thanks.