Skip to main content

Face Detection and Face Recognition

 
A developer at Apple made a comment on Twitter about iOS5 including Face Detection technology.  He said very clearly that it was face detection, and in fact elaborated a little bit later (in a conversation about technology in MacOS) that it was face detection and not face recognition. 

This was picked up by the blog 9to5Mac, and spread from there to other blogs.  9to5Mac reported correctly that the tweet referred to face detection, but then the story switched into a discussion of face recognition in iOS5.  Soon everyone was writing that iOS5 would have face recognition technology built-in. A few of these articles are here, here and here, but there are lots more. (Note, by the way, that some writers got the distinction correctly, such as this article.)

The point is that face detection and face recognition are two very different technologies.  Face detection is finding faces in a picture, like when your digital camera draws a square around faces and tries to focus on them. Face recognition is seeing whose face is in the picture.

Yes, it's true that Apple acquired Polar Rose last September, and Polar Rose has some very strong face recognition technology.  So it makes perfect sense to speculate that Apple will come out with face recognition in iOS.  But a tweet about face detection doesn't say anything about face recognition.

It's also true that face recognition on mobile is a hot area that's likely (in my opinion) to explode by the end of 2011.  Google's recent acquisition of PittPatt is only the latest of several face recognition companies that Google has acquired over the years.  Face.com has strong technology in the area, and Facebook is in the process of rolling out increasingly strong versions of face recognition.  Possible applications on cellphones include phone security (instead of a password), auto-tagging people in pictures before posting them to Facebook, sending pictures automatically to people whose faces are in the pictures, and much more. 

So yes, face recognition is a really hot area, and yes, it's likely that Apple will bring it to iOS and Google will bring it into Android.  Apple may even bring it to iOS5. But even when we're reading tweets it's important to read carefully.  Face detection is exciting enough, and will undoubtedly lead to lots of great apps.

Popular posts from this blog

Sensor-fusion indoor location system in over 30 malls in Singapore

SingTel just deployed indoor mapping and navigation from Swedish start-up  SenionLab  in over 30 shopping malls in Singapore, with 20 more in planning, in the biggest single deployment we've seen yet of customized indoor location technology. The most significant aspect of this deployment is that this is one of the first commercial deployments of indoor location technology that uses a new technology approach called "sensor fusion," which uses sensors in smartphones (gyroscope, compass, accelerometer, ...) to track location by sensing the phone's movements. Most indoor location systems determine location by measuring the signals of nearby Wi-Fi hotspots or cellular antennas, but this risks looking choppy as the phone waits for the radio signals to change to detect the next place it is. Sensing motion using phone sensors makes the system work more smoothly, regardless of radio signal details.  (The SenionLab solution does use Wi-Fi signals as well, to compe...
Read more

See great indoor location tech from the 2017 testbed

The videos from the Indoor Location Testbed at GeoIoT World 2017 are now released! Here is your chance to see how these solutions performed in our real-world evaluation, in the videos below. The testbed evaluated each solution by walking around a real-world venue, the GeoIoT World 2017 conference, measuring performance at 10 pre-selected points. Click here for the testbed report, which analyzes each solution's performance in a wide variety of metrics, including real-time accuracy, accuracy stabilization, consistency, latency, floor change, first fix, setup time & more. Let's start with BlooLoc's tag-based solution, which achieved accuracy under 2m in real-time and under 1.5m after stabilization: Then let's look at the infrastructure-free solution from GipsTech , which achieved accuracy under 2m without using any beacons or radio signals: Next is GipsTech's solution with BLE added: Next is BlooLoc's phone-based solution: Fin...
Read more

Seeing Quuppa's indoor location technology at MWC 2015

I first met Fabio and Kimmo from Quuppa in 2012 , before Quuppa existed, when they had spent years researching indoor location positioning technologies at Nokia Research Center. Less than a year after that they formed Quuppa , independent of Nokia, to bring their technology to market. At the time I called Quuppa " the newest and oldest in indoor location positioning ." Fast forward two and a half years. Their technology has reached market, it can now track both smartphones and BLE devices, and it's accurate to within 20-50cm. Before reading more about their technology, take a look at a video of their demo  in action: Grizzly Analytics has analyzed indoor location technologies by over 150 companies , and virtually all of the radio-based technologies operate by measuring the distance  between the device being tracked and other radio devices, and using these distance measurements for either multilateration or fingerprinting. For example, the well known BLE beacons me...
Read more