mobile btn
Wednesday, November 29th, 2023

Homeland Security S&T awards nearly $200,000 to startup Flux Tensor for motion-identifying algorithms

© Shutterstock

The startup Flux Tensor captured the interest of the United States Department of Homeland Security (DHS) Science and Technology Directorate (S&T) last week with the potential to apply motion-identifying algorithms to complex security videos and improve safety for soft targets.

The department will provide a $199,833 Phase 1 Other Transaction award to the Overland Park, Kansas-based company through its Securing Soft Targets solicitation. S&T, as part of its Silicon Valley Innovation Program (SVIP), has sought ways to automatically detect anomalous events through camera feeds, reduce error from human performance and simultaneously reduce delays and increase responsiveness during threat situations. It views the algorithms proposed here as having potential for schools, sports arenas, transportation systems, shopping centers, places of worship, and other public venues.

“Monitoring motion in soft targets is difficult in large areas where changes in light and weather can impact video quality,” Melissa Oh, managing director of SVIP, said. “Technologies that enhance real-time monitoring in situations where visibility is low due to environmental constraints will improve responders’ ability to maintain security in and around soft target venues.”

Flux Tensor created a flexible object detection means of identifying what it called persistent change, supposedly capable of factoring in weather changes, light variations, or poor video quality in near real-time while detecting objects in motion. Specifically, DHS believes this could identify motion of interest – and potential danger.

The software is in development but could help reduce the burden left by limited checkpoints for screening passengers and their belongings, at the very least. Ideally, DHS sees potential use for mass transit and mass gatherings, providing added layers of security without impacting travel times.