Will ever be available again this feature for programatic AI usage?
I am building a code for a project and prior to November I was using this feature for object recognition in an image. It was really accurate and now it is deprecated … do not really know why.
How ever the third party options as Google Vision, AWS Rekognition or Microsoft Azutre are not even close for the purpose what I was using it for.
Welcome to the community!
Have you tried using the 4-turbo or the 4o series of models?
Yes I tried but no luck did a detailed research and this is came across …
" OpenAI has deprecated direct image processing through its API, so if you’re using openai.ChatCompletion
or other similar methods for image-related tasks, they will no longer work in openai>=1.0.0
. However, there are still some ways to achieve object identification in images with OpenAI and other alternatives"
I did setup a google vision APi and tested it … works OK but by far is not what I need. The google vision when is text recognition response is based on what ever text is displayed in the image … what is OK but is not what I am looking for.
My applicatin captures images with 3 cameras and sends it up to AI for identification.
For example in the image it is a Tre Stelle - Fetra Cheese container what I really need is : Brand: Tre Stelle, Category: Dairy, Type: Feta Cheese nothing else.
The Google Vision gives me this but in a really random way due to the fact that I am sending up 3 independent images and also all the other writing is recognised what is not needed.
The original request what I had to OpenAI was - identify object and retunr Brand: Category and Type: was this simple and got exactly what I needed.
I am really new to the Google Vision it could be that I can have this type of filter configured just I do not know yet how to get it …