Contoso web app with Azure OpenAi: input images

Question

Hello,

I used AI foundry to build a private chatGPT on our cloud (with Azure OpenAI). The default interface is with Contoso, and it is really easy to implement. However, it does not accept image as input, only text. The model I used (GPT-4o) is multi modal and accepts images if I test it on AI foundry. But on the deployed interface (Contoso), it does not accept images. Is there a way to activate this option? Or another simple interface?

Thank you very much for your help,

Best

Answer 1

Hi Desiree Petrilli,
Greetings & Welcome to the Microsoft Q&A forum! Thank you for sharing your query.

It sounds like you're using a vision-enabled model like GPT-4o, which can indeed process images. However, the Contoso web app interface might not support image input by default. Here are a few steps you can take to enable image input or find an alternative interface:

You can customize the Contoso web app to support image uploads by modifying the front-end and updating the backend to send images to the GPT-4o model using the OpenAI API’s vision endpoint. This requires some coding to handle image uploads and format them correctly for API requests.
Alternatively, you can build a simple interface using tools like Streamlit, Flask, or React that allows users to upload images and send them to Azure OpenAI. The request should include the image in base64 format, following OpenAI’s guidelines for multimodal inputs.

I hope this information helps.