Inside The Technology Behind Facial Recognition | NBC News Now

Inside The Technology Behind Facial Recognition | NBC News Now

Posted by


BIASED, ESPECIALLY WHEN IT COMES TO DETECTING MINORITIES. TO DETECTING MINORITIES. WERE THE LATEST ON WHERE IT’S WERE THE LATEST ON WHERE IT’S BEEN AND HOW IT WORKS. BEEN AND HOW IT WORKS.>>RIGHT NOW WE HAVE THREE>>RIGHT NOW WE HAVE THREE REALLY GOOD EXAMPLES OF FACE REALLY GOOD EXAMPLES OF FACE RECOGNITION BEING BANNED WITH A RECOGNITION BEING BANNED WITH A SPECIFIC CITY. SPECIFIC CITY. SAN FRANCISCO IS ONE OF THEM. SAN FRANCISCO IS ONE OF THEM. ANOTHER IS OAKLAND. ANOTHER IS OAKLAND. THE THIRD IS SUMMERVILLE IN THE THIRD IS SUMMERVILLE IN MASSACHUSETTS. MASSACHUSETTS. THERE’S ALSO A BILL BEING THERE’S ALSO A BILL BEING DISCUSSED RIGHT NOW FOR THE DISCUSSED RIGHT NOW FOR THE STATE OF MICHIGAN. STATE OF MICHIGAN. THE FIRST STATEWIDE. THE FIRST STATEWIDE. WE’VE LEARNED THAT CITIES IN WE’VE LEARNED THAT CITIES IN MICHIGAN INCLUDING DETROIT HAVE MICHIGAN INCLUDING DETROIT HAVE HAD IT DEPLOYED FOR YEARS. HAD IT DEPLOYED FOR YEARS. THIS IS AN IMPORTANT MOMENT FOR THIS IS AN IMPORTANT MOMENT FOR THE U.S. THE U.S. IT IS THE FIRST TIME WE’RE IT IS THE FIRST TIME WE’RE SEEING CHALLENGES ABOUT SEEING CHALLENGES ABOUT SURVEILLANCE AND BIOMETRIC DATA SURVEILLANCE AND BIOMETRIC DATA BEING USED FOR PURPOSES OF LAW BEING USED FOR PURPOSES OF LAW ENFORCEMENT. ENFORCEMENT. BEING TRANSFERRED INTO THE BEING TRANSFERRED INTO THE COURTS. COURTS.>>TODAY WE’RE GETTING THIS ON>>TODAY WE’RE GETTING THIS ON FACIAL RECOGNITION. FACIAL RECOGNITION. WHEN AN ARTIFICIALLY WHEN AN ARTIFICIALLY INTELLIGENCE PROGRAM SEES A INTELLIGENCE PROGRAM SEES A HUMAN FACE LIKE THIS, TELL US HUMAN FACE LIKE THIS, TELL US WHAT IT’S HOOK FOR. WHAT IT’S HOOK FOR.>>WELL, THERE ARE LOTS OF>>WELL, THERE ARE LOTS OF METHODOLOGIES. METHODOLOGIES. ONE IS CALLED OPEN SPACE. ONE IS CALLED OPEN SPACE. THE FIRST IS TO FIND OUT WHETHER THE FIRST IS TO FIND OUT WHETHER THERE’S EVEN A FACE IN THE THERE’S EVEN A FACE IN THE PHOTO. PHOTO. IT DOES IT BY LOOKING FOR 58 IT DOES IT BY LOOKING FOR 58 DIFFERENT FACIAL LAND MARKS THAT DIFFERENT FACIAL LAND MARKS THAT TYPICALLY REPRESENT WHAT WE TYPICALLY REPRESENT WHAT WE WOULD THINK OF AS VISIBLY HUMAN WOULD THINK OF AS VISIBLY HUMAN FACE. FACE. WITHIN THOSE WE’RE LOOKING FOR WITHIN THOSE WE’RE LOOKING FOR EDGES. EDGES. THE EDGES REPRESENT A GROUP OF THE EDGES REPRESENT A GROUP OF PIXELS THAT ARE THE SAME COLOR PIXELS THAT ARE THE SAME COLOR OR SHIFT FROM A DARKER COLOR TO OR SHIFT FROM A DARKER COLOR TO A LIGHTER ONE. A LIGHTER ONE. IN THE OPEN FACE PAPER, THERE IS IN THE OPEN FACE PAPER, THERE IS PARTICULARLY EIGHT DIFFERENT PARTICULARLY EIGHT DIFFERENT LAND MARKS THAT REPRESENT A LAND MARKS THAT REPRESENT A HUMAN EYE AND MANY OTHERS HUMAN EYE AND MANY OTHERS COMPRISE AN EYEBROW OR A MOUTH COMPRISE AN EYEBROW OR A MOUTH OR A NOSE. OR A NOSE. THE THINGS YOU MIGHT SEE WHEN THE THINGS YOU MIGHT SEE WHEN YOU LOOK IN A MIRROR. YOU LOOK IN A MIRROR.>>NOW THAT WE KNOW WE HAVE A>>NOW THAT WE KNOW WE HAVE A HUMAN FACE, THE GOAL IS TO GET HUMAN FACE, THE GOAL IS TO GET IT INTO A FORMAT THAT’S UNIFORM. IT INTO A FORMAT THAT’S UNIFORM. WE CAN SCALE OR ROTATE OR EVEN WE CAN SCALE OR ROTATE OR EVEN ADJUST THE ANGLE OF THE ADJUST THE ANGLE OF THE INDIVIDUAL FACE. INDIVIDUAL FACE. THIS WILL HELP IT UNDERSTAND. THIS WILL HELP IT UNDERSTAND.>>WE WANT EVERY FACE WE’RE>>WE WANT EVERY FACE WE’RE WORKING ON TO BE PREDICTABLE WORKING ON TO BE PREDICTABLE ENOUGH. ENOUGH. THE CHALLENGE COMES IN WHEN YOUR THE CHALLENGE COMES IN WHEN YOUR TRAINING DATA. TRAINING DATA. YOU HAVE A REALLY SUNNY DAY, IT YOU HAVE A REALLY SUNNY DAY, IT WILL BE REALLY DIFFICULT TO WILL BE REALLY DIFFICULT TO PREDICT. PREDICT. IF IT IS SOMETHING AS IF IT IS SOMETHING AS PREDICTABLE AS A PAIR OF PREDICTABLE AS A PAIR OF SUNGLASSES. SUNGLASSES. THE NEXT STEP IS TO TRY ON THE NEXT STEP IS TO TRY ON EXTRACT INFORMATION FROM THAT EXTRACT INFORMATION FROM THAT FACE. FACE. WE DO THAT BYPASSING THE PHOTO WE DO THAT BYPASSING THE PHOTO THROUGH A NEURAL NET. THROUGH A NEURAL NET. EACH ONE WILL TRY TO EXTRACT EACH ONE WILL TRY TO EXTRACT FEATURES AND THEN REPRESENT THEM FEATURES AND THEN REPRESENT THEM IN A MATHEMATICAL FORM. IN A MATHEMATICAL FORM.>>WHERE ARE THESE PROGRAMS>>WHERE ARE THESE PROGRAMS GETTING THE DATA TO MATCH THE GETTING THE DATA TO MATCH THE FACES? FACES?>>USUALLY A FACIAL RECOGNITION>>USUALLY A FACIAL RECOGNITION SYSTEM HAS A DATABASE OF PHOTOS SYSTEM HAS A DATABASE OF PHOTOS THAT HAVE BEEN RUN THROUGH. THAT HAVE BEEN RUN THROUGH. THEY’RE ALREADY TRANSLATED INTO THEY’RE ALREADY TRANSLATED INTO MAPS. MAPS. IF I HAVE A DATABASE AND THAT I IF I HAVE A DATABASE AND THAT I KNOW THIS MATHEMATICAL KNOW THIS MATHEMATICAL REPRESENTATION LOOKS LIKE LIZ REPRESENTATION LOOKS LIKE LIZ O’SULLIVAN, I CAN TAKE THIS NEW O’SULLIVAN, I CAN TAKE THIS NEW PHOTO AND TRY TO FIGURE OUT HOW PHOTO AND TRY TO FIGURE OUT HOW CLOSE THAT VECTOR SPACE IS TO CLOSE THAT VECTOR SPACE IS TO THE RESULTING VECTOR SPACE OF MY THE RESULTING VECTOR SPACE OF MY NEW PHOTO. NEW PHOTO. IF IT IS VERY CLOSE, THEN YES. IF IT IS VERY CLOSE, THEN YES. IT MIGHT BE LIZ. IT MIGHT BE LIZ. IF THEY’RE VERY FAR APART, IT IS IF THEY’RE VERY FAR APART, IT IS LIKELY SOMEONE ELSE. LIKELY SOMEONE ELSE.>>PEOPLE ARE STARTING TO>>PEOPLE ARE STARTING TO REALIZE HOW POWERFUL THIS TOOL REALIZE HOW POWERFUL THIS TOOL IS. IS. LAW ENFORCEMENT AND FEDERAL LAW ENFORCEMENT AND FEDERAL AGENCIES HAVE BEEN USING IT. AGENCIES HAVE BEEN USING IT. IN A LOT OF CASES THEY’RE IN A LOT OF CASES THEY’RE FINDING IT DOESN’T WORK AS WELL FINDING IT DOESN’T WORK AS WELL ON PEOPLE OF COLOR OR WOMEN OR ON PEOPLE OF COLOR OR WOMEN OR NONBINARY PEOPLE. NONBINARY PEOPLE. SO THIS IS IMPORTANT FOR US TO SO THIS IS IMPORTANT FOR US TO TAKE A MOMENT TO UNDERSTAND TAKE A MOMENT TO UNDERSTAND FULLY EXACTLY WHAT WE WANT OUR

4 comments

  1. Not something I or many want. Big Brother is already in our individual lives too much and people hand it over, clueless.. Social Media's… "see yourself older" b.s. etc.

  2. I retired in New Mexico. The state DOT implemented facial scanning in order to get a driver's license. I refused. They refused to renew my license. I've been without a driver's license now for a little over ten years. It causes a few inconveniences but nothing major. I live in an area with good mass transit so I sold my truck. I purchased a nice touring bicycle with saddle bags for runs to the grocer. State government here doesn't (yet) demand citizens carry a current form of photo ID. But if I lived in say Texas or several other states I'd be arrested and tossed in jail. For nothing more than refusing to give government personal bio metric data. This has to stop somewhere. Government at all levels in this country has become arrogant in its attitude towards citizens. They literally behave as if they own the citizenry. They don't – at least not me – and I'll resist them to my dying breath.

Leave a Reply

Your email address will not be published. Required fields are marked *