Microsoft's new http://how-old.net/ website: impressive facial recognition

[Aristos Queue]

Microsoft has launched a new website. At http://how-old.net/# , you can upload a photo and the site tells you how old the person in the image is. I've been playing with various photos. The thing is extremely accurate as far as I can tell. I'm impressed with how far facial recognition has come.

[MarkCG]

Microsoft has launched a new website. At http://how-old.net/# , you can upload a photo and the site tells you how old the person in the image is. I've been playing with various photos. The thing is extremely accurate as far as I can tell. I'm impressed with how far facial recognition has come.

 

It tells me I look 47. I blame the Actor Framework for turning my hair gray

[infinitenothing]

Over by 5. I'm still impressed.

[MikaelH]

-2 years, not bad at all

[jcarmody]

Nice try, NSA.

[hooovahh]

Well I went to my facebook and I could only find one face shot of me that was detected (not many pictures of me on my facebook I guess).  In the photo was me and my wife.  It thought I was 13 years older than I was when the photo was taken, and it thought my wife was 19 years older than she was.

[jcarmody]

 

It thought I was 13 years older than I was when the photo was taken, and it thought my wife was 19 years older than she was.

 

You're working too hard.

[drjdpowell]

It thought I was 13 years older than I was when the photo was taken,

Ha!!  Only thought I was 11 years older.  

[hooovahh]

Oh my, so I think I'm going to stop trying.  I just went and used my linkedin profile picture, and it thought I was 24 years older than I was when the picture was taken.  Of course I was wearing a hat, not that I know how that affects it.  Seriously it thought I was in my 50s. 

[eberaud]

+4, not bad

[mje]

My profile picture puts me bang on for when it was taken. Frightening and awesome.

[Yair]

Microsoft has launched a new website. At http://how-old.net/# , you can upload a photo and the site tells you how old the person in the image is. I've been playing with various photos. The thing is extremely accurate as far as I can tell. I'm impressed with how far facial recognition has come.

 

They have a link there now describing the background, which may not have been there originally. Essentially, this was a demo meant to show the face-rec functions in Azure, and I'm guessing the age part was just a nice gimmick they added which uses machine learning, not something actually designed to be super accurate. The basic takeaway is that you can use the APIs yourself for doing similar things (although I never looked at the technical side of Azure, so I don't know what the requirements for it are and if it's callable from LV).

[ShaunR]

I don't know what the requirements for it are and if it's callable from LV).

Its a REST API.

[Yair]

Its a REST API.

 

Ah, so it would be callable from LV easily enough, but would require a web connection (not surprising). So I guess that means that Azure is less like .NET and maybe more like .NET for the cloud.

[ShaunR]

Ah, so it would be callable from LV easily enough, but would require a web connection (not surprising). So I guess that means that Azure is less like .NET and maybe more like .NET for the cloud.

 

Not really. It's a service.

This is the direction of all applications. The current trend is for your Computer, Gamebox, whatever to merely be a terminal - a bit like in the old Unix days - with minimal software in the form of a client. That way they have a cross platform solution that does not expose their valuable intellectual property at all (closed source with no distribution-monopolize the technology), can completely control access (control over copyright), ensures reliance on their services (control over downstream developers and services) and hoover up all your information to sell.

Edited May 3, 2015 by ShaunR

[Aristos Queue]

Not really. It's a service.

This is the direction of all applications. The current trend is for your Computer, Gamebox, whatever to merely be a terminal - a bit like in the old Unix days - with minimal software in the form of a client. That way they have a cross platform solution that does not expose their valuable intellectual property at all (closed source with no distribution-monopolize the technology), can completely control access (control over copyright), ensures reliance on their services (control over downstream developers and services) and hoover up all your information to sell.

There's a few of us at NI -- in several different products -- that saw this trend a couple years ago when Adobe first started trying it out; we preemptively raised objections to LabVIEW moving in that direction, even though there were absolutely no plans to do so at the time. I'm glad we did... because although I can imagine NI will have some products at some point that follow this model, but I do not expect LabVIEW to go that way. Speaking as an R&D insider: I observe enough of an "allergy" within NI to LV adopting this model that it is unlikely to happen.

 

Now, taking off my R&D hat and putting on my LV user hat: I don't like the software as service model. I get why businesses like it though. If you also don't like the software-as-only-a-service model, please mention your objections to your local Field Sales agent, as a ward against future ideas. There are a lot of advantages to *customers* to the software-as-only-service model, and Adobe was able to make the change because it got the customers who liked the portability/upgradability/no-IT-involvement-ability of the service model excited about it first and closed down a lot of the objections before they even got raised. By the time most Adobe users even heard about the change, it was fait accompli, with blog posts from users excited by the change holding top search result slots. So you might want to occasionally mention to Field Sales, "You know, I really like owning my own software." Express not just objections but also advantages with the current situation. Doing so will help keep the allergy strong!

[ShaunR]

There's a few of us at NI -- in several different products -- that saw this trend a couple years ago when Adobe first started trying it out; we preemptively raised objections to LabVIEW moving in that direction, even though there were absolutely no plans to do so at the time. I'm glad we did... because although I can imagine NI will have some products at some point that follow this model, but I do not expect LabVIEW to go that way. Speaking as an R&D insider: I observe enough of an "allergy" within NI to LV adopting this model that it is unlikely to happen.

 

Now, taking off my R&D hat and putting on my LV user hat: I don't like the software as service model. I get why businesses like it though. If you also don't like the software-as-only-a-service model, please mention your objections to your local Field Sales agent, as a ward against future ideas. There are a lot of advantages to *customers* to the software-as-only-service model, and Adobe was able to make the change because it got the customers who liked the portability/upgradability/no-IT-involvement-ability of the service model excited about it first and closed down a lot of the objections before they even got raised. By the time most Adobe users even heard about the change, it was fait accompli, with blog posts from users excited by the change holding top search result slots. So you might want to occasionally mention to Field Sales, "You know, I really like owning my own software." Express not just objections but also advantages with the current situation. Doing so will help keep the allergy strong!

 

I agree absolutely. In fact. I see it as a parasitic business model. The problem is there are very few arguable reasons why you should not uses these services  and even less for why businesses should not adopt them. The one clear winner against is privacy and security but there seems to be a general malaise towards that.

 

It is getting a huge boost from the games industries who are perfecting the technologies and normalising expectations.  There is little "skin in the game" in that sector for customers and although some are moaning about not being able to play during outages and not "owning" software; there is general acceptance due to demand for multilayer online games which require internet connections anyway.

 

To a lesser extent, Bitcoin is leading the way too even in the face of huge security failures which can be squarely aimed at online services in isolation and not Bitcoin itself. People seem willing to give away their money to faceless wallet websites when they could just as easily have it far more securely on their own computers and phones.

 

The saving grace for NI, i suppose, is that LabVIEW programmers tend to work in industries where software can amputate limbs so no service business wants responsibility for that and no customer wants such machines connected to the internet.

Edited May 3, 2015 by ShaunR

[mje]

...and Adobe was able to make the change because it got the customers who liked the portability/upgradability/no-IT-involvement-ability of the service model excited about it first and closed down a lot of the objections before they even got raised.

 

I for one really liked some of their products, but can't ever see myself giving them money again so long as they stay exclusively on this "cloud" platform. I don't want access to my data to be held hostage for $x/mo indefinitely, never mind the performance considerations for pushing large amounts of data around the cloud.

 

For what it's worth, NI would become irrelevant overnight for me if it went to a cloud platform. There are just so many reasons not to do it, especially in a corporate landscape.

[shoneill]

This is the direction of all applications

 

Oh good lord I hope you're wrong.

 

I could always start training to be a carpenter.  I always wanted to be a carpenter.

Edited May 4, 2015 by shoneill

[hooovahh]

 ...hoover up all your information to sell.

I prefer Dirt Devils in my house, and reserve "hoovering" to positive things like presidents...and dams.

 

And for me I guess I'd have a fall back career as an electrician.  If software I guess something in CAN, maybe develop my own CAN hardware and software package.  Ehh I'll just retire, I never liked work anyway.

[Yair]

... because although I can imagine NI will have some products at some point that follow this model, but I do not expect LabVIEW to go that way.

 

NI already has this - the LV web UI builder is one example. FPGA compile servers are another (although that's essentially just processing). I think this model has its place, particularly in providing some muscle to web apps, which I assume is part of the intention of Azure.

 

I certainly agree that I wouldn't want to see LV moving in that direction, and I don't think it's very likely either, but I would like it if LV did adopt one feature from this model - frequent updates. It would be nice if instead of having to install incompatible LV versions once or twice a year to get features and bugfixes I could just apply auto-updating patches which would be released at much smaller intervals and so only need to actually install a new version every 3-5 years or so.

[smithd]

I also imagine people in favor of this model would be happy to mistake liking a piece of software vs liking a style of sw. For example for a long time my group has used yEd for diagramming things. yEd is fine, but its kind of irritating to use and last year I found a similar tool online (https://www.draw.io/) which is significantly better in terms of usability and a little bit better in terms of features. But I'd really love it if I could just download it and use it offline.

[ShaunR]

 last year I found a similar tool online (https://www.draw.io/) which is significantly better in terms of usability and a little bit better in terms of features.

 

Oh dear. Now I'm conflicted.. I'd like to see something similar to that for designing new LabVIEW controls and would put up with an online solution to get that feature (although the compiled control would have to be downloadable to the local drive).

Edited May 4, 2015 by ShaunR

[hooovahh]

yEd is fine, but its kind of irritating to use and last year I found a similar tool online (https://www.draw.io/) which is significantly better in terms of usability 

Wow that is impressive.  But like you said I would be equally impressed with it if I could download it.  But of course then there is less control, and they'd likely charge for it.  Today they don't mind making it free because if they want to charge for it then they will and all users will be forced to pay.  If they allowed you to download it version 1.0 could be free and version 2.0 could be paid.  But then users would just continue to use version 1.0 until there was a good reason to upgrade.  I could see this business model working well no wonder companies are considering cloud based applications more.

 

Oh and my favorite piece of software I wish I could download is the Infinite Jukebox.  Ever get a song in your head you couldn't get out?  Try listening it for hours with no obvious start or end.

[crelf]

Oh and my favorite piece of software I wish I could download is the Infinite Jukebox.  Ever get a song in your head you couldn't get out?  Try listening it for hours with no obvious start or end.

 

That site is addicting. Although it seems to stop playing tracks if the window is minimized, or if I go to a different browser tab.

 

This almost blew my mind: http://labs.echonest.com/Uploader/index.html?trid=TRQDXXM13AFAB66B3F