Carrot is a model built by Plaintain Labs and hosted on Banana.dev The inspiration came from seeing general purpose text models like GPT3 and asking "can we do this for computer vision too?"
Hey! You can change the imageURL in the modelParameters to any image url you want to use, or better would be signing up for the api key to get unlimited use
@derekpankaew the model has some OCR capabilities which has impressed us (especially the math example), but the OCR isn't consistent enough where I'd lean on it for financial matters
Hi all ๐!
Makers Kyle / Sahil / Erik here! We are ML developers from Banana.dev that spun out into an ML research lab: PlantainLabs
We were inspired by general purpose text-based models like GPT3 that launched over a year ago and asked ourselves: Can we do something like this for computer vision?
So we launched Carrot!
With carrot you can generate state-of-the-art captions of images, ask questions to images, calculate image-text similarity, and more.
Try it and let us know what you think!
Cool thing! I admire every product made with GPT-3, it's such a magic and complex thing! Guys, ale you open for integrations? In digitalfirst.ai were looking for a AI text generator to integrate right now, we'd be also interested in train our own models in the future. If you're open for a quick talk let me know! (lena@digitalfirst.ai)
Grapevine