Client:
· A team of neuroscientists at a well-known research university, in Chicago.
· Their research objective:Investigate dysphagia or swallowing disorders in humans caused by deteriorating neurological and brain conditions.
· They have shot 1000’s of videos of rodents and prime mates, recoding their chewing and swallowing behavior under variety of conditions, by surgically implanting markers in the mouth
· These are high resolution, high speed X-ray videos, 250 frames a second, shot with two sets of cameras to create a 3D image.
Challenge
· Marker detection and tracking, across frames in 3D. The current process is manual labor intensive, expensive, error prone and not scalable.
· NSF gave them a target – further research to be funded, this process needs to speeded up, at least 300 times.
Action:
· They signed up for our Jumpstart Program. Our R&D lab became their R&D lab
· We got the videos in our lab. Our expert team started working with the researchers.
· We developed a set of machine vision and video analytics algorithms using deep learning techniques to identify the markers, and track them across frames in 3D.
Results
· With high accuracy we were able to speed up the process over 500 times.
· Our client has re-started the research funding process.
