Word Camera

Wow. I just took this photo: and it detected that I was wearing a hat and headphones, weaving those into the story. Impressive. Read it here.

"word.camera uses convolutional neural networks (via Clarifai) to extract concept words from images. It expands those initial words (mostly nouns) into sentences and paragraphs using a lexical relationship database (ConceptNet) and a flexible template system."

tried this the other day too - is there a service out there can can just pull out the words and attach tags automatically?

Really interesting results. Could be a great starting point for content creation. Upload an image of something you want to write about, then edit and add things in your own voice. The results are thorough, but don't sound human.

Hi all, I'm Ross Goodwin and word.camera is my project. I'm a grad student at NYU ITP, and I made this web app for a class I'm taking—I'm thrilled that it has received so much attention. I'm here to answer questions if anyone has any.