Pose Search

Lead Participant: EATON MEDIA LIMITED

Abstract

The goal of this project is to develop a robust machine-learning search tool that operates on human poses and locates exact or similar poses across a large dataset of photography and figurative artwork. The tool will allow users to either search for 'similar poses' to a current reference image, or to pose a virtual 3d mannequin and have that pose and camera view entered into the search.

The tool will integrate the current state of the art in pose-estimation algorithms along with a custom annotated dataset of dynamic photographic and artwork from BodiesinMotion.photo. The tool will give end users poweful ans precise image search for human figures, in a way that is not currently possible with text and keyword search.

Lead Participant

Project Cost

Grant Offer

 

Participant

EATON MEDIA LIMITED

Publications

10 25 50