Pictures at an exposition

Here’s another prototype along the path to automatic image recognition: ALIPR (Automatic Linguistic Indexing of Pictures in Real-Time) will take an uploaded picture or URL, and come back with certain words it thinks describe that image.  The results are pretty mixed right now, but ALIPR claims to be able to learn, if you will select the best tags from its list and add ones it missed.  This is a long way from automatic ALT text, but there’s a lot of good potential accessibility in approaches like this.

Speaking of learning, there’s no indication whether ALIPR is already studying up on the millions of web images that do have good ALT text.

By the way, here’s a cool app that their site offers right now: write or upload a short story, and ALIPR will suggest images to illustrate it with.

Photography: ALIPR Learns How to Auto-Tag Photos

This entry was posted in Blind, Communication, Education, Entertainment, Information management, Just a prototype so far, Just plain cool, Promising component, Redundant output, Software. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *