Facial Recognition

          As more and more people start using social media, it is only natural that technology should evolve to better suit the users. For example, Facebook managed to endure the test of time by adapting to its ever-changing audience and implementing new features while retaining older ones. One feature that is currently being developed is a feature called DeepFace, a facial recognition feature.

          Even though facial detection and (very) basic facial recognition has already been implemented into Facebook, DeepFace promises to be "as accurate as a human being at a few constrained facial recognition tasks". Essentially this program will allow the identification of faces in any of the 400 million new photos that are being uploaded daily.

          DeepFace can prove to be a very controversial system, especially in the realms of privacy. On one hand it can help to alert you when a photo is uploaded without your permission. The director of Facebook's artificial intelligence research, Yann LeCun, explains that this can actually help to bolster privacy in that after receiving an alert from Facebook telling you that you appear in a photo, you can then choose to blur out your face.

          Yet on the other hand, it will still trouble many, myself included, with the thought of being recognized at all. If DeepFace were limited to tagging me instantly on my friend's photos (or my friend-of-a-friend's photos) like the current system on Facebook is doing, I would be okay with it. In fact I may even welcome it. Yet the thought of being tagged in some stranger's photo - just that thought of being recognized by a piece of code - leaves me feeling a bit uneasy even if it gives me the option to blur out my face.

          In addition, the more photos we upload, the better the program will be at recognizing us. One of the techniques used to develop programs like those in facial recognition is called deep learning where the system will use a set of training data as a model for higher-leveled abstractions. For example, a facial recognition program can simply scrap through a library of celebrities' faces and use them as a training exercise in order to learn how to identify other faces with this technique. Seeing how Facebook's library of tagged photos is growing drastically each day, there is no doubt that DeepFace has a near endless amount of training data.

          This brings in the question of how consent will play into DeepFace. DeepFace uses a library called Social Face Classifaction (SFC) for deep learning. Facebook users must give permission for their personal data to be used when they sign up for the website, and this can potentially mean uploading their personal data into the SFC database. The extent of DeepFace using our private data can have many repercussions. For one, Facebook has already confirmed that law enforcement authorities use facial recognition programs to search through our photos (when compelled by a court order). But soon we may find ourselves with facial recognition technology advanced enough to allow advertisers to follow us whenever we are on a camera.

          Being recognized anywhere at anytime may become possible with this sort of technology. I for one would be thoroughly creeped out if I can instantly be recognized by anyone carrying around a computer with such programs. Imagine if you are walking down a street and a random person can just look at you and find out your name, and any other data that can pulled from social media or even criminal databases. Because of this I'm certain that facial recognition will have huge impacts on future privacy laws.