*** Facial recognition algorithms are challenged in identifying faces partially concealed by masks, the National Institute of Standards of Technology found. While the lower success rate won't come as much of a surprise, NIST's study also reported that that false matches are also reduced by masks. NIST was able to test recognition prowess on datasets of 6.2 million images of 1 million individuals using 89 different algorithms. The most accurate algorithms posted failure rates of 5% against masked targets but NIST found that some algorithms notched 20% to 50% failure rates.
*** The Federal Emergency Management Agency wants to buy an off the shelf public alert and warning tool to try out at one of its testing facilities in suburban Washington, D.C. The system, it said in a July 23 request for information, will provide emergency managers at FEMA's Indian Head, Md. Integrated Public Alert and Warning System app with 24/7 capabilities to send or cancel an emergency alert and warning message on behalf of State, Local, Tribal, and Territorial emergency managers. The acquisition, said the agency, will help determine if an off-the-shelf can contact federal officials who can then issue federal-level public alerts and emergency warning emergencies.
Posted on Jul 28, 2020 at 2:38 AM