This repository contains resources for an Amazon SageMaker workshop . Content is based on below notebooks and has been tweaked to make it work in a workshop format!
- From Unlabeled Data to a Deployed Machine Learning Model: A SageMaker Ground Truth Demonstration for Object Detection
- TensorFlow 2 Complete Project Workflow in Amazon SageMaker
Before starting with this workshop ensure following requirements are met:
- Create an S3 bucket with a cors policy. See this link on how to configure the CORS policy
- Create a SageMaker Studio Domain, make sure the execution policy has read and write access to the created bucket
- If you are running this workshop in a single AWS account make sure following service quotas are raised via a support request
per workshop participant you will need:
- One ml.p2.xlarge , ml.c5.xlarge, ml.c5.2xlarge and Total Instance Count for SageMaker Training
- Two ml.m5.xlarge and Total Instance Count for SageMaker Processing
- One ml.c5.xlarge, ml.m4.xlarge and Total Instance Count for SageMaker Batch Transform
- One ml.m5.xlarge and Total Instance Count for SageMaker Endpoints (Hosting)
- Two KernelGateway-ml.t3.medium for SageMaker Studio
- One Max User Profile per Account for SageMaker Studio
- Log into your AWS console
- Navigate to SageMaker --> Amazon SageMaker Studio
- If you don't have a user profile yet, create a user profile via "Add user" , make sure you select the execution role created during the prerequisites
- Once the profile is created, select open studio
- this can take about 5 minutes the first time
- Open a system terminal
- Clone this repository via
git clone https://github.com/johanneslanger/sagemaker-workshop-with-ground-truth
- In the file browser on the left open following notebook
sagemaker-workshop-with-ground-truth/ground_truth_object_detection_tutorial/object_detection_tutorial.ipynb
- When prompted select the Kernel `Python 3 (Data Science)``
- You can monitor Kernel status in the bottom left corner
- Follow the instructions of the notebook. Please note that the Kernel needs about 2-3 minutes to start the first time!
! Note only one workshop participant should create a private workteam for the labeling job.
! The labeling job should only be kicked off by a single participant.
! The rest of the notebook can be done by all workshop particpiants!
- Make sure you have cloned the repository to your SageMaker Studio environment as shown in lab 1.
- Then open following notebook and follow the instructions
sagemaker-workshop-with-ground-truth/tf-2-workflow-smpipelines/tf-2-workflow-smpipelines.ipynb
- As Kernel select Python 3 (Tensor_Flow 2.3 Python 3.7 CPU Optimized)
The contents of this repository are licensed ander the Apache 2.0 License except where otherwise noted.