OpenStack 04/09/2015 (p.m.)

• Facebook’s Deepface Software Has Gotten Them in Deep Trouble | nsnbc international

  Tags: surveillance state, Facebook, litigation, facial-recognition, biometrics

  • In a Chicago court, several Facebook users filed a class-action lawsuit against the social media giant for allegedly violating its users’ privacy rights to acquire the largest privately held stash of biometric face-recognition data in the world.

    The court documents reveal claims that “Facebook began violating the Illinois Biometric Information Privacy Act (IBIPA) of 2008 in 2010, in a purported attempt to make the process of tagging friends easier.”

  • This was accomplished through the “tag suggestions” feature provided by Facebook which “scans all pictures uploaded by users and identifies any Facebook friends they may want to tag.”

    The Facebook users maintain that this feature is a “form of data mining [that] violates user’s privacy”.

    One plaintiff said this is a “brazen disregard for its users’ privacy rights,” through which Facebook has “secretly amassed the world’s largest privately held database of consumer biometrics data.”

    Because “Facebook actively conceals” their protocol using “faceprint databases” to identify Facebook users in photos, and “doesn’t disclose its wholesale biometrics data collection practices in its privacy policies, nor does it even ask users to acknowledge them.”

  • This would be a violation of the IBIPA which states it is “unlawful to collect biometric data without written notice to the subject stating the purpose and length of the data collection, and without obtaining the subject’s written release.”

    Because all users are automatically part of the “faceprint’ facial recognition program, this is an illegal act in the state of Illinois, according to the complaint.

    Jay Edelson, attorney for the plaintiffs, asserts the opt-out ability to prevent other Facebook users from tagging them in photos is “insufficient”.

  • Deepface is the name of the new technology researchers at Facebook created in order to identify people in pictures; mimicking the way humans recognize the differences in each other’s faces.

    Facebook has already implemented facial recognition software (FRS) to suggest names for tagging photos; however Deepface can “identify faces from a side view” as well as when the person is directly facing the camera in the picture.

    In 2013, Erin Egan, chief privacy officer for Facebook, said that this upgrade “would give users better control over their personal information, by making it easier to identify posted photos in which they appear.”

    Egan explained: “Our goal is to facilitate tagging so that people know when there are photos of them on our service.”

    Facebook has stated that they retain information from their users that is syphoned from all across the web. This data is used to increase Facebook’s profits with the information being sold for marketing purposes.

    This is the impressive feature of Deepface; as previous FRS can only decipher faces in images that are frontal views of people.

    Shockingly, Deepface displays 97.25% accuracy in identifying faces in photos. That is quite a feat considering humans have a 97.53% accuracy rate.

    In order to ensure accuracy, Deepface “conducts its analysis based on more than 120 million different parameters.”

Posted from Diigo. The rest of Open Web group favorite links are here.