Facial Recognition and Online Privacy

I haven’t been blogging for a long time, mainly du to a change of industry but it also since I have had less time to share my thoughts about technology. There was though one post that I started to write on in february and I decided to post this weekend. It is about privacy online and the development of what can be argued to be integrity intruding technologies, in this case the development of facial and pattern recognition technologies.

There was several articles in february that I planned to comment regarding people being tracked down online, but so much time has passed so I don’t even remember my point anymore.What I do remember, which might be old information now, is my thoughts around facial recognition and the recent development in these area that we have seen. What facinates me with these technology developments is the value the bring and the potential to change behavior online. That technologies can change normal usage is of course nothing new, but there is an aspect of it when it could as well have a negative effect on usage. Facial recognition is not new, but the fact that it is becoming available for everyone is an interesting development. That means the ability to find and recognise anyone online. Image search today is very limited and pictures are only found if the have been tagged or contains the used keyword in the name or description, which of course seldom is the case. This is the problem a growing number of facial recognition companies are aiming to solve.

Polar Rose is a browser plugin that lets you discover who is who in any public photo. Photos are tagged by using facial recognition and by assigning a tag to a photo, other photos with that person can be recognised.

Picollator has been in beta mode since january and is a multimedia search and indexing engine. It scans Internet, collects digital pictures and makes the index, matching visual objects in the images by pattern recognition. You can simply upload a photo with people to launch the search process.

TinEye was launched recently and is an image identification-based web search engine, that doesn’t do any facial or pattern recognition, instead it enables you to find matching pictures. This one I have heard was useful on flickr in a case when someone suspected his pictures were used without permission.

There has been reports about Google trying out facial recognition where the tag face is added to show pictures with a face, so it still offers very limited functionality. Reuters has also been testing facial recognition in video search. Another player is Riya. They started out with facial recognition and visual search, but then moved into object recognition by launching Like.com that allows you to find similar looking products.

Among these facial recognition products – Riya is trying a business model where they utilize their technology for ecommerce. Polarrose I don’t know, their technology is open-source (which of course is great) but as a browser plugin I cannot understand how they will make money from it. Maybe from the integration into different services such as flickr and facebook so the tagging becomes better or similar. Technologies like Picollator (part of recogmission.com) is a technology that is possible to integrate in to search engines, beefing up the existing image search.

There is though an interesting issue coming up, and that is the fact that what is put up on the web is often available to the public. In the good old time, users have lived in a pleasant belief that what is put up will most likely not be found since the web is so vast –  now with better search engines and better pattern recognition services everything can be found easily, a face on a party picture, a presence in a video clip etc. It is just like being tagged on facebook but it covers the whole web and it is automatic. If someone just once have said a person is you, or use a picture of you, they will be able to find you in every media from pictures to videos.

How will this work with copyright and integrity when a previously anonymous person you took a photo of on the street suddenly has a name and identity? Could have interesting complications….

I have always promoted open communities since I wanted what I share to be accessible for friends without the need of being members. But with a technology such as Polarrose offering something like facebook tagging for everything that is public there might be a move towards more closed sharing where you decide what you want public and what you want to only be accessible for friends and family. Projects like open social and single sign on will of course make closed networks easier to access for friends without the need of additional memberships.

So apart from the integrity issue when it comes to widely deployed facial recognition and tagging I do look forward to pattern recognition that enables you to find similair looking objects and photos. That could be very useful when you for example want to find more photos of a car – you can just search for similar ones.

Interesting applications of facial recognition is the two entertainment services provided by Optical Recognition Obectives; Play-Analogia for finding celebrities lookalike and IconDate that is a dating site where you can find a date that looks like a certain celebrity you chose or any other person you want your date to look like. That has to be value adding for dating 🙂

The reason I came into the privacy issue was due to an experience in february at an event in Shanghai. It was an afterwork and one guy walked around taking photos, encouraged by the organiser and the people who posed. But there were also a few people starting to question it and didn’t allow him to take photos of them, asking if they would be put up on facebook. It was clear that they were worried about having the pictures tagged on facebook. I talked with the photographer and heard that it happened at a house-party in Sweden as well, people started to talk about their legal rights and that he is not allowed to take photos of them. An interesting development of how social networks is affecting peoples real life behavior such as photographing a at a party. Which of course is good since it is related. Swedish organisations have published official warnings to people urging them not to tag or comment friends on Facebook due to the implications it can have at a later stage job interview etc.

Despite the privacy implications the discussed technologies have it is not necessary bad. The online behaviour today is not always normal and well thought through, and a change in it will be good for people that for example put up less suitable pictures of themself without realising they easily can be found. At the same time are laws sometimes not adopted to the changing society in all and we have seen this regarding filesharing, music and movies…..maybe we will now enter into the discussion of sharing private information.

~ by Andreas Sigurdsson on September 2, 2008.

One Response to “Facial Recognition and Online Privacy”

  1. Andreas, thank you for the review. I am quite sure that our photos are our traces or faceprints in the web. And we do not know how this information can be used without our permission.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: