24e0082987928d56e9088dd40c8804e0.jpeg

TrueFace.AI is here to catch the facial recognition tricksters


TrueFace.AI knows if it's looking at a real face or just a photo of one.
Image: ian waldie/Getty Images

Facial recognition technology is more prevalent than ever before. It’s being used to identify people in airports, put a stop to child sex trafficking, and shame jaywalkers.

But the technology isn’t perfect. One major flaw: It sometimes can’t tell the difference between a living person’s face and a photo of that person held up in front of a scanner.

TrueFace.AI facial recognition is trying to fix that flaw. Launched on Product Hunt in June, it’s meant to detect “picture attacks.”

The company originally created Chui in 2014 to work with customized smart homes. Then they realized clients were using it more for security purposes, and TrueFace.AI was born.

Shaun Moore, one of the creators of TrueFace.AI, gave us some more insight into the technology.

“We saw an opportunity to expand our reach further and support use cases from ATM identity verification to access control for data centers,” said Moore. “The only way we could reach scale across industries would be by stripping out the core tech and building a platform that allows anyone to use the technology we developed.”

“We knew we had to focus on spoof detection and how we could lower false positives.”

TrueFace.AI can detect when a face or multiple faces are present in a frame and get 68 raw points for facial recognition. But its more unique feature is spoof detection, which can tell real faces from photos.

“While working on our hardware, we tested and used every major facial recognition provider. We believe that doing that (testing every solution available) and applying facial recognition to a very hard use case, like access control and the smart home, allowed us to make a better, more applicable solution,” said Moore. “All of these steps led us to understand how we could effectively deploy technology like ours in a commercial environment.”

They made their final product by using deep learning. They trained classifiers with thousands of attack examples they collected over the years, and liked the results.

A “freemium” package is available to encourage the development community that helped TrueFace.AI come up with a solution. Beyond that, the Startup Package is $99 per month while the Scale Package is $199 per month. An Enterprise Plan is available via a custom agreement with TrueFace.AI.

While Moore couldn’t divulge exactly which companies are using the technology, he did say some of them are in the banking, telecommunications, and health care industries.

It’s a service that could become increasingly valuable as companies turn to facial recognition technology.

Read more: http://mashable.com/2017/07/07/trueface-ai-facial-recognition-photo-attack-detection/

3ab8905d060f1ce6f382f77c94a9b856.jpeg

Apple working on dedicated AI chip for iOS devices, report says

Image: lili sams/mashable

Apple’s reportedly working on a new kind of chip potentially for future iOS devices that’ll be used just for processing AI, Bloomberg reports.

Bloomberg says the chip’s called the Apple Neural Engine internally, and could be used for “offloading facial recognition in the photos application, some parts of speech recognition, and the iPhones predictive keyboard to the chip.”

By moving AI processing to a dedicated chip, battery life in devices could also see a boost since the main CPU and GPU wouldn’t be crunching as much data and gobbling as much power.

The report says Apple plans to integrate the chip into its devices, but it’s unclear when that’ll happen, and if any iOS devices launching this year will have it.

Apple’s work on an AI chip shouldn’t surprise anyone who’s paying attention to the competition. Virtually every tech company is working on improving AI processing on mobile devices.

Qualcomm’s latest Snapdragon 835 chip, which is already in devices like the Samsung Galaxy S8, has a special module dedicated to processing AI tasks.

Years ago, Apple started designing its own mobile processors to improve performance and reduce power consumption, and it’s really paid off.

Despite having fewer cores, the iPhone 7 still crushes the Galaxy S8 when it comes to sheer performance.

iPhones and iPads also come with an Apple-designed “M”-branded motion coprocessor to collect sensor data from the various included sensors (accelerometer, gyroscope, compass, etc.). It’s this M chip that helps with tracking health and fitness data.

Furthermore, in addition to the main Intel processors in the new MacBook Pros, there’s also a small Apple-made “T1” chip for powering the Touch Bar. Apple’s AirPods also have a custom W1 chip that helps with pairing them to iOS devices.

Clearly, Apple loves making custom chips for things. We’re all for it, especially if that means longer battery life.

If the future is AI everywhere (and it definitely looks like that’s where things are headed), it’s in Apple’s best interests to control the stack (like it always does) with its own AI chip.

Read more: http://mashable.com/2017/05/26/apple-artificial-intelligence-chip-iphone/