DIGTIS Training Using Docker Container

Hi Everyone,

I am new to this and following the, ‘Two Days to a Demo’ path. I am on the following step:

Issues posted with photos here: https://github.com/dusty-nv/jetson-inference/issues/202

So far I have successfully created the AWS P3 instance and have downloaded the data set. I have also created the sub folder mentioned that is title 12_classes. I cannot get DIGITS to train this. It tells me that it did not find two valid categories. I don’t necessarily believe this error due to it being a docker container. I am concerned how I am launching the docker container is causing my issue.

Docker Command:
docker run --shm-size=1g --ulimit memlock=-1 --runtime=nvidia --name digits -d -it -p 5000:5000 -v /home/ubuntu/efs/Data/12_classes:/efs/Data/12_classes nvidia/digits

There is another example docker command I could run, but when I try alternatives I cannot access the public IP for DIGITS running on the AWS P3 as a docker container.

Can anyone help me with what might be the error? I believe it is possibly that I don’t have a target directory configured for digits on my P3 instance? The data looks correct to me.

Resolved - https://github.com/dusty-nv/jetson-inference/issues/202