Conduent Patents Facial Expression Recognition IP for Multiple Industries

New technology considers the eyes as the proverbial 'windows to the soul,' as some people have described the way the eyes can be read to determine a person’s frame of mind.


When it comes to new-gen facial recognition software for identification and security purposes, the eyes definitely have it.

Conduent, Xerox’s former services business that split off and went public in January and brings in upwards of $6 billion in yearly revenue, has been awarded a U.S. patent for facial recognition intellectual property that gets most of its data and semantic information from a person’s eyes.

How does this work? Well, it in effect considers the eyes as the proverbial “windows to the soul,” as some people have described the way the eyes can be read to determine a person’s frame of mind.

The Conduent intellectual property locates faces within a video stream and automatically identifies a person's eyes to better understand facial expressions. Applying machine learning and pattern recognition techniques to that region of the image allows the facial expression to be classified from ubiquitous low-resolution surveillance grade video, the company said.

Using IP to Branch Out to New Use Cases

This automated facial expression recognition IP uses Conduent's expertise in computer vision techniques and is now enabling the company to pursue commercial opportunities in multiple industries, Chief Information Officer Carol Kline told eWEEK.

“We can use standard security video—no need for high-fidelity video for what our tech does—to analyze the region around your eyes, to be able to pattern satisfaction or emotion,” Kline said.

This can be valuable for brick-and-mortar retailers as they watch customers for such emotions as joy, delight, disgust or something else while they walk the aisles looking for what they want to buy. It can also help store owners and managers to become more optimal about how they present merchandise on shelves and in specific locations around the store.

It also can be used to gauge commuter satisfaction on public transit and improving accuracy of screenings in border-crossing control, Kline said.

Use cases for the facial recognition software in more detail:

In retail: Knowing shopper satisfaction at a more granular level allows the retailer to optimize the physical shopping experience including layout, merchandise and the checkout process. This technology, which can also read customer satisfaction during customer-employee interactions, creates an advantage for omni-channel retailers looking to enhance information gathered from their brick and mortar stores.

In transportation: Facial recognition can be similarly used to gauge rider satisfaction on public transit or customer satisfaction in customer-agent interactions. Through this technology, public transit agencies gain better insights to provide safer, cleaner and more courteous public transportation, further motivating commuters to use these services, including end-to-end multi-modal travel.

In border crossing control: The technology can cross-reference expressions of any persons identified by law enforcement with FBI and Homeland Security databases that scan across millions of photos in state and federal databases, as well as facial images collected from non-citizens as they enter the United States. This advancement could result in more screening for high risk individuals and potentially less for lower risk individuals, making border security more accurate, effective and efficient.

In vehicle passenger detection systems in HOV (high-occupancy vehicle) lanes: The system uses advanced video analytics to detect the number of occupants in moving vehicles with 95 percent accuracy at speeds up to 100 mph. Enforcing the rules of these high-occupancy lanes improves the customer experience for those who abide by the rules and helps governments validate the integrity of HOV lanes and High Occupancy Toll (HOT) lanes.

The patent, along with several others awarded this year, adds to the Florham Park, N.J.-based company’s portfolio of more than 300 U.S. patents and about 400 pending U.S. patent applications.

Chris Preimesberger

Chris J. Preimesberger

Chris J. Preimesberger is Editor-in-Chief of eWEEK and responsible for all the publication's coverage. In his 15 years and more than 4,000 articles at eWEEK, he has distinguished himself in reporting...