I2T: Image Parsing to Text Generation

diagram_semanticweb

Here is an example:

event_parsing_result

This would make it possible to search for images and videos in a new way like. “Guy wearing a blue jeans while snowboarding”. 🙂

Awesome!

[ source ]