Need autonomous driving training data? ›

Conversations in Machine Learning: Photo Storage & Sharing Goes Retro, but Better

Conversations in Machine Learning: Photo Storage & Sharing Goes Retro, but Better

This is another installment of Mighty AI’s “Conversations in Machine Learning” blog series. Each week, our content human, Cassie, shares a summary of a recent conversation we had with a machine learning team and potential customer—what they’re building, how they’re handling training data today, etc. Read more about the series here.

Hi there. How ya feeling? Good? Good. Time to geek out on cool machine learning stuff.

This week I’m highlighting a call we had with the makers of a consumer mobile application that is killing it in terms of downloads, usage, and even business model. You’ve quite likely heard of the app, but due to my vow to not be creepy or shady, I won’t name it. But I will talk about it…

So first, do you remember a time when people were not snapping selfies, photo-documenting their brunch, or capturing touching moments with a freaking iPad that blocks everyone else’s view? I know. It’s hard to recall such a time. We’re now sufficiently photo-obsessed. And that goes beyond the U.S.

Now, there are plenty of good photo storage options out there for all those precious snapshots, but there are a couple major flaws with them: 1) you gotta be at least a little technologically inclined to use them, and 2) many of them aren’t “smart,” so your photos just kinda get dumped into a virtual box with little meaningful organization. This app we’re talking about here today solves both of those problems and beyond, in an attempt to make your digital photography experience more like the scrapbooking days of yore.

How so? Well first of all it automagically aggregates pics from all over your digital life—storage, camera roll, social media, etc. You wouldn’t keep them all separate if you were taking Polaroids; doesn’t make a lot of sense for them to be spread out across devices and services. So thanks, anonymous app! Also: as photos are added, they’re intelligently organized, based not only on tags and timestamps (yawn), but also the contents of a photo, the setting in which it was taken—even what’s happening in it.

And aha! There’s the machine learning: computer vision. Sophisticated computer vision at that. And sophistication in machine learning = excellent training data, which, ahem, is probably why they’re talking to Mighty AI.

Right now they’ve got user-generated tags to work with (which can be unreliable in terms of accuracy), as well annotations from the image labeling they’re doing internally (which can be a time-suck). What they really need is a high volume of high-quality annotations on which to gauge the validity of their existing models. They need ground truth data to determine if their models are categorizing images correctly (and if not, what they need to tweak). And they need this training-data generation to be low-effort.

Mighty AI is very clearly in a strong position to help them with this—it’s, uh, exactly what we do. Lots of accurate annotations quickly and efficiently, with no project management, QAing, or really anything on the part of the customer? Yep. That’s our TDaaS™ workflow.

Pretty sure it would be bueno for everyone for our awesome tasking community to crank out these annotations but we’ll see, friends, we’ll see.

Until next time, we’ll be sharing relevant things on Twitter and LinkedIn, so catch us there.

image credit: Annie Spratt via CC0 1.0

Note: Prior to January 10, 2017, Mighty AI was known as Spare5. While Spare5 remains the name of our consumer brand and application, we’ve relaunched our business-customer side as Mighty AI, which also serves as the parent company under which Spare5 now lives. Some posts on have been updated with the new company name to ease confusion.