Skip to main content
Version: v1.5.17

Tutorial 4A: Importing and interacting with images

Overview​

This tutorial explores the workflow for importing images to a Collection so that you can ask questions about it later. To understand the workflow, we will explore a sample image of a medical invoice.

Objectives​

  • Understand the process of importing images into a Collection using Enterprise h2oGPTe.
  • Learn how to interact with and extract information from the imported image using the Chat feature in Enterprise h2oGPTe.

Prerequisites​

Step 1: Create a Collection​

Let's create a Collection that can store the image we want to explore.

  1. In the Enterprise h2oGPTe navigation menu, click Collections.
  2. Click + New collection.
  3. In the Collection name box, enter the following:
    Image: Medical Invoice
  4. Click + Create.

Step 2: Add the image to the Collection​

Now that we have a Collection let's add the downloaded image.

  1. Click + Add documents.
  2. In the + Add documents list, select Upload documents.
  3. Optional: In the OCR model list, select an Optical Character Recognition (OCR) model. The OCR model can identify and extract text from images and PDFs.
    note

    An OCR converts documents to text, including the auto-detection of the language on every page in the imported document.

  4. Click Browse files... and add the medical_invoice.jpg file.
  5. Click Add.
note
  • Enterprise h2oGPTe converts the imported image files into a PDF file.
  • The imported image is a JPG file, but Enterprise h2oGPTe can handle various other file types. To learn more, see Supported file types for a Collection.

Step 3: Chat with the Collection (image)​

Now, with the Collection containing the image, let's explore it.

  1. Click Start your first chat.
  2. In the Ask anything box, enter the following questions:
    What is this invoice about? Who issued it to whom? And what are the important amounts mentioned? 
  3. Click Submit.

Questions

Summary​

In this tutorial, we learned how to import an image into a Collection. To understand the workflow, we used an image of a medical invoice issued by Jefferson Healthcare.


Feedback