Multi-Source Multi-Scale Counting in Extremely Dense Crowd Images
UCF Office of Research & Commercialization, FL, United States
A software algorithm that produces nearly accurate crowd counts from video or still images containing an average of 1,280 people per frame. The algorithm functions with new constraints in multi-scale Markov random field to infer a single count over the entire image.
Primary Application Area: Electronics, Sensors & Communications
Technology Development Status: Proven Manufacturability
Technology Readiness Level: TRL 4
FIGURES OF MERIT
Value Proposition: Existing crowd-counting algorithms cannot distinguish individuals in crowds of hundreds or thousands, resulting in counting errors. The new invention leverages multiple sources of information to compute a more accurate estimate of the number of individuals present in a dense crowd visible from a single image.
Organization Type: Academic/Gov Lab
Showcase Booth #: 523
GOVT/EXTERNAL FUNDING SOURCES
Government Funding/Support to Date:
Primary Sources of Funding: Federal Grant
Looking for: Development / License Partners