In plain English
A vision model cuts an image into patches and uses attention to find shapes, text and relations. It reads text, describes scenes and reads charts well. It can fail on counting, fine detail and poor photos. Ask specific questions, check with simple rules and test on real messy images.
- Patch
- A small square cut from an image.
- OCR
- Reading text that appears in an image.
- Vision model
- A model that can take images as input.
Picture it
From picture to answer
1
Your image
Photo, scan, chart or screenshot
2
Cut into patches