Ever trained for a marathon?
I’m about to start, but not the kind you’re thinking of. IBM Watson is going to train me for the last season of Game of Thrones (after I train it, of course)
I, like everyone in the world, it seems, love Game of Thrones. I haven’t read the books or participated in the many, many online discussions, blogs, live tweet frenzies, etc., that accompany any running season of the series. I’m just entertained by it. Therein lies my problem. Half the time, I have no idea what is going on, let alone who is who. I get it, for the most part, but in Game of Thrones fashion, they’ll flash to someone who last appeared in act 3 of the 6th episode of season 2. While there’s often an audible gasp at my viewing party, I’ll have no idea who they are and will have to, as is most often the case, end up googling. I will certainly understand their significance after reading any of the thousand posts per second that are generated by a given episode of Game of Thrones.
I want to fix that! So, for the next year, in advance of the final season of Game of Thrones, I will use technology to do my bidding. By that, I mean that I will use Watson. Yes, IBM’s Jeopardy ass-kicking Watson, which is now a set of Cognitive Computing services that anyone, including my GoT novice-self, can spin up and get working for them. I will train a visual recognition application to identify Game of Thrones characters. If the penultimate season was any indication, this last season is going to be chock-full of revelations, new (really old, but new(ish) to me) faces and the meeting of powerful forces and characters.
Let the training begin.
Within a few minutes, I provisioned a free trial on IBM Cloud, a platform comprised of applications, infrastructure, and cognitive services, powered by Watson.
For this project, I’ll use Visual Recognition, a deep learning image classifier.
Let’s start easy. Everyone knows who Arya is. A girl has no name, sword-wielding, kill list reciting, kick-ass Stark sister, who is brought to life by a carpool karaoke master. Maisie, if you’re reading this (as if she’d be reading this), give me a shout out. I want to play the snap game!
We start by training Watson with a collection of Arya images.
As you can see, Watson identifies her quite easily.
Now, for a hard one. Let’s be clear. Hard for me. I’m sure a GoT aficionado reading this will think this far too easy an example, but hey, we (Watson and I) have to start somewhere, right? Beric Dondarrion. I mean, who is this guy? I vaguely remember him from several seasons back, but his significance requires some hard core GoT brain picking.
Let’s take Watson for a spin.
There you have it. Watson understands the contents of the images and correctly classifies this dude as Beric. Seriously, who is he?
While I’m using a visual recognition service for my GoT needs, this technology has real life (I know, I know, this is real life!) applications and implications.
Saving the planet, for example. OmniEarth leveraged the Watson Visual Recognition cognitive service to analyze their “extensive collection of aerial images of drought-stricken lots for specific areas where water usage could be scaled back”. As a result, specific recommendations could be made to property owners in drought areas. This data-driven approach will help usher in the “green data revolution”.
At IBM, my role is that of Technical Evangelist. That means that I get to play with technology, and teach people what is possible, how it can be leveraged, and most importantly, how they can build and expand their businesses with it.
As part of this role, I get to collaborate with a group of equally passionate and, let’s be honest, nerdy evangelists, one of whom is Erica Reuter. We’ve decided to record a podcast that blends together two of our passions: Tech and TV. The premiere episode of Unsupervised Binge Watching, where we discuss, with a tech twist, The Sopranos, and Game of Thrones, can be found here.
Have a great idea for the use of visual recognition in your organization? Want to embed it within your app? Develop IP around it? Let’s connect. To take IBM Cloud for a spin yourself, sign up for free here.