Last night at the monthly Tokyo 2.0 event, Google Mobile Product Managers Riku Inoue & Brad Ellis from Google Japan, gave a presentation entitled Convergence and beyond.
In my opinion the most interesting part was during the Q&A session that followed, I asked if Google had plans to add people or face recognition to Google Goggles? This would be possible since: Picassa has face recognition, Flickr has added people tagging, and of course Facebook has support for tagging people in photos.
Riku and Brad’s response was Google has person recognition working in a Google internal build of Goggles, but the feature was removed in the publicly released version due to privacy concerns.
However they also said, Google hasn’t given up on the feature and are trying to find the right approach to make sure people’s privacy is taken into account. When they do, the People recognition feature will be added to the public version of Goggles.
The 14 min. Google Japan presentation can be watched below in full (presentation is in english and Japanese. Apologies if the video quality isn’t good at times, this was recorded with my N97.)
Google Japan Presents Search Trends across platforms at Tokyo 2.0 February.
If you would rather read what was presented, here’s my overview of the highlights:
The presentation started with Brad Ellis talking about web trends across mobile and PC. The presentation included (sanitized) data of Google services use throughout an average week, a week with a holiday, and a week day.
The data shows some interesting trends, like when people go on Lunch break both mobile web and PC web use increases. By looking at the graph you can see when the majority of people leave their office to go home, PC web usage drops while mobile web use increases. Also their data shows that on weekends and holidays Mobile web usage peaks.
Next Riku Inoue talked about mobile convergence, saying ‘something magical happens’ when rich hardware devices have an always on connection to the cloud. It enables innovative new services to be created.
He said the iPad is a very good example of mobile and PC converging. (Wonder if this was a hint of things to come for a Chrome OS tablet device?)
As two examples of some of the possibilities that cloud-connected mobile devices provide, Google Goggles and Voice Search on the Android HTC Magic (aka HT-03A in Japan) were demonstrated to applause from the audience. During the demo a Japanese female audience member loudly commented the Android mobile looks like the iPhone, to which one of the product managers replied this is not an iPhone.
To wrap up the APIs for Google Voice recognition were mentioned that are available now for developers to use. Riku said, for example there are some developers working on a twitter app where you can say something and it will tweet what you said.
I for one am looking forward to more easily remembering people’s names with Google Goggles, as you may have guessed yes I’m bad with names. Also the speak to tweet application sounds like it has potential for being useful when driving or when your hands are busy. This would be similar to what Ford Sync announced at CES, Microsoft Sync and Ford are also working on speak to tweet functionality for cars.
I've been anticipating face recognition functionality from Google Goggles since it came out. The privacy concerns would be readily allayed by a process in which participants could opt-in via their Google profile and even use their linked profile photo and even Picasa-tagged images to immediately source Google's recognition engine.
In the same vein, I am very much hoping for Shazam app-style voice recognition, which would allow the user to identify who is speaking via sampling their voice, a technology we've had a long time and is likely due for iPhone-level use. Would be useful during public speaking events and other places where you can hear people but have no idea who is talking.
How do you plan to use people recognition to remember people's names (like you mentioned)? Go up to the person and take a photo with your phone? Or do you have some other setup in mind?
I thought the speech to text bit was done (and available on the Nexus 1). Any place where you can type, you can also speak (as long as you have decent enough internet access).
Goggles on my dev1 managed to identify a couple of (very large) personalities. It even identified one by a painting, so yep, if people opt in and put their Picasa data online, I'm sure Goggles'd manage with that as well.
Is the unofficial version available on the net?
In some “dark” channels?
A file name to search?
Google is waiting for a scapegoat which offers a similar service. The first will be defeated.
How Google will restrict the service?
Will Google offer the photos on a Google server?
That would be shit!
It must be possible to determine the identity of a person (if the person or someone else left the data).
Imagine:
A policeman beats down a protester / pedestrians.
Or he enters a person lying on the ground.
Police and prosecutors will not pursue this.
Or the Verfahtren is set. Or are there only a small fine.
Possibly, the state protects the identity of the perpetrator.
The offender remains on duty. This is Germany.
With the name, criminal / civil action may be brought.
I correct:
Police and prosecutors will not pursue this.
>>>Or criminal proceedings are terminated.<<< Or are there only a small fine.