There was a great Twitter thread that explains how to use GPT-4 and Visual-ChatGPT to generate food recipes from a photo of a fridge. It also suggests combining the Visual-ChatGPT API from Huggingface with GPT-4 API and using more powerful Visual Question Answering models like Viper to get better outputs.

This person fed the image model to Visual-ChatGPT and it asked to describe in detail what was in the fridge.

"The image you provided is a refrigerator with food items such as apples, oranges, bananas, eggs, a carton of milk, cheese, and a package of lunch meat and other items."

There are lots of very valid responses from the AI and I think incorporating a feature like this to “scan” my fridge or scan my receipt / upload a picture of each with the AI built into the app would be very beneficial and give similar outcomes. However, if possible, I would like for all of it to be within one app so you don’t have to go to Chat-GPT as well as use Shelf Life.