Galleries

Back to: Galleries // 2018 Spark:Experience // IBM PowerAI Vision

<< Previous Entry    Next Entry >>

IBM PowerAI Vision

Winner - Gold

Competition: Spark:Experience
Designer: Natalie Miller - IBM Cognitive Systems Design Manager
Design Type: Web application
Company / Organization / School: IBM Cognitive Systems
Team Members: Natalie Miller, Tim O'Keefe, Gabi Campagna, Michael Davis, Kylee Barnard, Srinivas Chitiveli & Mike Hollinger

It can take years of study for an oncologist to learn to identify certain cancel cells.  It is difficult to monitor real time video of an airport to determine if a bag is left unattended.   Although AI algorithms and their speed have exponentially improved, only data scientists with special training and knowledge are able to make use of it.  Additionally, the algorithms can require tens of thousands of accurately labeled images or video during its training phase. PowerAI Vision is designed to enable domain experts who work with images or video, use their expertise to explore and build AI models which are useful in their field.  It provides a complete workflow with defaults ensuring immediate success.  Users draw boxes around objects within images using a simple drawing tool.  After 10 boxes are drawn, the software automatically learns and begins drawing boxes around likely objects which the user then verifies.  When enough images are labeled, the user clicks a button to train the model and can watch a graph as training occurs. When training completes, the user can test the model by just dragging and dropping an image.