June 18, 2012. Facebook acquires a Tel Aviv-based company called Face.com, which had developed a facial recognition platform to identify images on web and mobile applications. The Israeli company was able to tag, before being acquired by the Zuckerberg Social Network, more than 18,000 million faces.
Almost at the same time as Face.com, Facebook buys FacioMetrics, which developed a system that applies advances in artificial intelligence (AI) to analyze images of faces in mobile applications, and even in the identification of the emotions they transmit . With the FacioMetrics software it is possible to know if the user is sad or happy when taking a picture and analyzing the different parameters of the human face.
FacioMetrics was created within Carnegie Mellon University, and developed computer vision and machine learning algorithms capable of detecting seven different emotions on people’s faces through their photos.
With these two acquisitions, Face.com and FacioMetrics, and some more like Masquerade, Facebook develops DeepFace, a facial recognition system combined with deep learning or deep learning. With DeepFace, which uses a 9-layer neural network and 120 million connections, Facebook is able to identify human faces in digital images. The system is 97% correct, compared to 85% for the image recognition system used by the FBI.
DeepFace not only connects a person’s face to their name, it can literally find a needle in a haystack and find a face in an image that contains thousands of different faces. According to experts, the human brain is only 28 percent more accurate than the facial recognition program developed by Facebook from the companies it acquired.
Facebook is far from the only company that has ventured to recognize its users with facial recognition. So do Apple, Google, Samsung. Microsoft or Huawei, for example. And what you have to ask yourself is what is the true purpose of requiring users to accept and authorize the recognition of their faces.
Facebook has already faced resistance to DeepFace. In Europe it was prohibited, but since mid-April users of the Social Network are asked if they want to authorize facial recognition when they open their feed. And from here the questions of the users arise, when the Cambridge Analytica scandal is still very recent, which has led Zuckerberg to the United States Congress to give explanations about the transfer of data from millions of customers to third companies.
Where is the data of the faces recognized by DeepFace stored? Who can access them? What does Facebook do with the images of its customers? Does facial recognition work only when Facebook is open or also when other applications are open? Could facial recognition be used to identify someone participating in a legal protest? Surely Facebook is capable of answering them and reassuring their customers, if they are not.
