Skip to Main Content
Use Data Labeling Service to Create a Biomedical Image Classification Model

About This Workshop

Youtube Video

About This Workshop
Biomedical image classification has become increasingly popular in recent years. Many physicians and research facilities are leveraging image classification to assist in classifying biomedical images into distinct categories. This technology can assist physicians in making diagnoses and can also be leveraged in research. Although this lab focuses on biomedical images, the use of the Oracle services highlighted in this lab can be applied across a wide range of industries and use cases.

This tutorial teaches end-users how to bulk upload images to object storage, create a dataset in the Data Labeling Service, bulk label the images, and create/train a custom image classification model using the labeled dataset. Each image contains a scan of a blood sample and is labeled with one of three classifications: cell, debris, stripe. The goal is to leverage DLS and OCI Vision to classify these images.

Workshop Info

1 hour, 30 minutes

Introduction

Get Started

Lab 1: Use DLS to Bulk Label Dataset

  • Policy Setup
  • Task 1: Create an Object Storage Bucket
  • Task 2: Upload the images from your local machine into your bucket
  • Task3: Create a Data Labeling Service Dataset
  • Task 4: Populate your DLS Dataset with the data from your Object Storage Bucket

Lab 2: Create Custom AI Vision Model

  • Policy Setup
  • Task 1: Create a Project
  • Task 2: Create a Custom Image Classification Model
  • Task 3: Train your Custom Model + Submit
  • Task 4: Test the Model on New Images
  • Assumes that the end-user has full-administrative privileges in the tenancy
  • Some familiarity with OCI CLI is desirable, but not required
  • Some familiarity with Python is desirable, but not required

Other Workshops you might like