Sending photo automatically to chat GPT

I am PhD student. I try to send images automatically to the chat GPT is it possible? do you have a suggestion?

Could you elaborate on what you’re trying to do here? What sort of images and for what purpose?

The developer forum is a great resource for learning about and working with OpenAI’s API, so perhaps a dedicated app/tool would be better than some system auto-pasting into ChatGPT.

In case you want the vision aspect of ChatGPT, you could read this documentation.

There you will learn how to post an image to the OpenAI API.

Keep in mind that this will cost money as the API is not free.

Best of luck! :hugs:

I’m working on a project where surveillance camera footage is analyzed by ChatGPT to generate reports. As a first step, I need an automated API that can send, for example, 100 images at a time to ChatGPT for processing. Do you have any suggestion?

What you intend to do is make a tool that can analyse recorded footage, right? If so, there’s a decent article on the OpenAI cookbook for processing videos with GPT-4o.

OpenAI’s API provides numerous models with vision capabilities, j.wischnat sent the relevant documentation for that earlier.

Are you a developer yourself, or are you new to this? Do keep in mind the API is a paid service and not as simple to use as a website like ChatGPT.

thank you for suggestion of this document, I am totally new in subject, I am a civil engineer. I Do not have an idea, my supervisor said that :frowning:

This can be a daunting task to start with if you’ve never done anything like it before!
I can’t spoonfeed you the actual solution but here are some tips on getting started:

Think of how you want to achieve your goal and how to minimise needing to learn something new:

Apply anything you know already to the project, even if it’s inefficient, it’s better to get things working than to be stuck learning.

A tip for videos is that you don’t usually need every single frame.
You could do every 2nd frame (which cuts down the requests you send in half) or even every 3rd frame (1/3 of the requests you’d send normally).
Of course you can go even lower depending on how much needs to be in context.

Let’s say you want to check if a parking spot is in use!

Send an image to OpenAI API asking whether a car is parked there → Get response → Next request can probably wait until 5-10 minutes later.

That way you don’t need to actually send a whole video where 99% of the frames are the same anyway.

Feel free to keep updating or let us know where you’re stuck so we can help! :hugs: