Real-time text search technique wins patent for face images lacking tags

A little-known startup in Arizona has been granted a U.S. patent for a neural network capable of searching facial images on live video feeds, in real time, using text queries and without data-labeling.

In its announcement of the patent, AlgoFace says its edge-based biometric facial analysis software enables searches without identifying a subject using still or video content from conventional color and near-infrared cameras.

The company has stated that its software is not a facial recognition technique and, in fact, it will “never enter any industry where a false positive or false negative from our artificial intelligence technologies can be used to rob a person of their human rights.”

In a commitment only good as long as AlgoFace owns the patented technique or remains an independent business, the firm says it “will never enter the facial recognition industry.”

Other software companies already are in the facial recognition search engine field, including Baidu, which was testing algorithms in 2012.

Among the proposed applications for the technique, according to the company, are useful face search engines and programmatic advertising.

As will virtually all advancements related to face biometrics, AlgoFace promises better, “unbiased” searches for lost or abducted children caught on live video. In this case, it would require typing “a basic description.”

The software can detect and track 209 facial landmarks, including foreheads and inner cheeks, according to the company. Each eye reportedly is assigned 50 landmarks. Accuracy is reportedly not affected by masks.

Strategic Advisor appointed

Jennifer Jordan has joined AlgoFace as a strategic advisor, with CEO Andrew Bart saying he expects her to challenge him on ethical AI-related topics, as well as provide connections and experience with early stage, AI, and capital markets to advance the company’s identity-free biometrics in the market.

Jordan helped build boutique Pacific Northwest investment bank Black & Company into a mid-sized institution, after which it was acquired, and also managed investor relations as part of a 10-year stint at Cadence Design Systems. She has also made direct venture capital investments in early-stage tech companies while with MassVentures. She brings experience from serving on the Board of Directors at several businesses and a pair of non-profits.

“AlgoFace is bringing to market the most inclusive dataset in the U.S. for facial analytics, while preserving consumer privacy and security,” Jordan comments in the announcement, referencing common concerns of biometric algorithm bias based on unbalanced training data. “The company offers commercial solutions leveraging its FaceTrace platform for cosmetics, medicine, automotive customers. I am excited to support the team in growing the company.”

Jordan is currently a managing director with Techstars for the NY Barclays Accelerator and Female Founders First program, as well as mentoring innovators through several organizations.

Article Topics

AI  |  Algoface  |  biometrics  |  biometrics at the edge  |  computer vision  |  ethics  |  facial analysis  |  patents  |  privacy  |  real-time biometrics