Label and train activity

This repo is the base structure of the AI workshop: https://github.com/Demo-AI-Edge-Crazy-Train/workshop-model-training.
The instructions of the workshops are available here: https://rivieradev2024-crazytrain.netlify.app/.

Content

annotations contains the datasets/traffic-sign-lego annotations
|_ annotations/annations.json is used to populate label studio with annotations and pictures of the dataset
|_ annotations/sink is the label used in training scripts and are generated by label studio
container-images are some of the Containerfile used

datasets contains the lego datasets taken from the train
|_ datasets/traffic-sign-lego is the dataset itself
|_ datasets/validation-dataset.tar.gz is used to test the lego model after traning

manifests contains the object storage, label studio and data science project deployment

operators contains the required operators

operators-instances contains the required operators instanaces

workshop [WORK IN PROGRESS] contains some scripts and indications to setup the workshop for a fixed number of participants (here 40)

Deployment

With the following commands you will deploy the base structure of the workshop. To setup the workshop for multiple participant, you will have to navigate to workshop directory (WIP to automate).

You will have to provision some data science project componenets on your own: data connection (use a https endpoint), pipeline server, workbench using a custom image (quay.io/demo-ai-edge-crazy-train/workbench:minimal) with the data connection well linked. Refer to the workshop instructions.

oc apply -k operators
# Wait for RHOAI operator installation
oc apply -k operators-instances
# Wait for RHOAI components deployment
oc kustomize --enable-helm manifests | oc apply -f -