Remark Holdings has scored extremely well in the latest round of the NIST’s Face Recognition Vendor Test (FRVT). In the test, the NIST evaluated different facial recognition systems to gauge their ability to identify people who are wearing masks, and the offering from the Remark AI business finished in the top five in a field with 198 competitors.
The solutions were ranked based on criteria like storage, memory, and matching speed and accuracy. The Remark offering performed well in all categories, and was good at recognizing faces covered with masks and at recognizing those without.
In that regard, Remark claimed that its solution is particularly good at matching faces in difficult circumstances. The company’s algorithm also earned a top 15 spot in a separate FRVT evaluation that assessed performance in wild unconstrained scenarios. In plain terms, that means that Remark could match faces at sharp angles, even when those faces were distorted, blurred, or captured in adverse lighting conditions.
“The results of the FRVT establish that Remark AI’s software system is a top solution, outperforming the systems of many billion-dollar unicorns,” said Remark Holdings Chairman and CEO Kai-Shing Tao. “Remark AI not only provides the highest level of accuracy, but also maintains the highest level of data privacy and protection on the market.”
According to Remark, potential applications of its computer vision technology include pedestrian monitoring, and event, action, and temperature detection. The solution can match faces in video footage, which gives it utility in a wide range of different industries. On that note, Remark is pitching its tech for those in the hospitality, healthcare, transportation, and manufacturing sectors, amongst several others.
Remark’s KanKan subsidiary supplied the Hangzhou Primary School System with a Smart Campus solution with facial recognition capabilities back in 2019. The Chinese city of Xi’an started using KanKan’s face tech to verify the identities of taxi drivers earlier that same year.
July 30, 2021 – by Eric Weiss