Recently I met with Jenny Maher, CEO of Coolivy, a body-measurement startup for Apple and Android devices. While the body-measurement space has been heating up for the last two years, Coolivy’s smartphone-based technology shows that companies are simplifying the process enough that mainstream consumer adoption of body-measurement tech is getting closer.
We started doing this in September of last year. In the beginning, we wanted to do a global apparel e-commerce website, a platform. But we realized there were a lot of issues that we can’t really control. But we still wanted to stay a platform, because I believe in an open platform – that’s where the creativity is. So we were thinking of a platform where patten designers could meet facilitators, so that eventually you could go from the idea to a garment, cutting out the middle-man.
But that would be really expensive. So we were thinking we wanted to make custom clothing. Then pretty soon we had another problem: cost. You have to send tailors to customers’ homes, because customers don’t feel confident in their own measurements, because they make mistakes and a lot can’t reach the places they need to measure.
That’s got to be a huge overhead.
There is. But the thing is, customers don’t feel confident in measuring themselves, or don’t know how to measure, and a lot of times, they don’t want to be responsible for the measurement. They could make a mistake and the garment could turn out wrong, and they don’t want to be responsible for that. And even if you send tailors to measure and they make the clothes, there’s still a pretty high rate of alteration. There’s always the chance of human error. This all together makes the costs very high.
Today, even though it’s Web 2.0, Web 3.0. you still have to stand up with a real human to measure people. So we started searching for the technology to measure people. In my research, there 2 or 3 types of solutions. One is the TSA style. It’s like a microwave; what it does is detect a body’s water content.
The second type looks more like a photo booth. It’s a box, so you have four pillars, four corners with cameras on each of them. They take photos from different angles and reconstruct it to view in 3D.
Both kinds are bulky, pretty big, booths. They’re expensive. Most stores can’t afford, or don’t want to buy them. I think they can sell it to shopping malls, but who’s going to foot the bill?
I don’t think they are the future. I think the future should be something portable, something everyone has. So there’s another type of solution with Xbox Kinect, like BodyMetrics. The user-base is much smaller than smartphones. They do have an advantage over our solution: the infrared camera. Infrared beams are sent to you and sent back so I can tell the distance. The good thing about that is, for one, I know what’s you, and what’s behind you is not you. So I’m measuring you. A smartphone doesn’t have an infrared camera, so we require that you are in front of a background, so that there’s nothing moving behind you.
To scan their bodies with Coolivy, users stand in front of the camera and turn 360 degrees. You do need a background (a wall, a door, a backdrop, etc.) that is 6+ feet tall, and 3+ feet wide, basically, a space that allows you to make a 360 degree turn. You can do the measurement yourself, so you don’t need a second person.
What’s the tech behind it?
It’s called optical flow. The whole idea is that video is numerous images — we analyze each image, taking the 2D point of view and projecting it into 3D space. Then we are able to reconstruct a 3D image of you.
We use the face as a reference, we reference the whole body, to know that it’s you and not something else like a chair or something. Then we separate the body from the background. We basically clean the body.
Compared to something like the face or feet, the human body is hard to measure. Facial recognition is already a very mature science. The relative positioning is clear – the eyes are here, the nose is here. For the feet, you have five toes. The body is the hardest to measure because it’s long, it doesn’t have edges.
How many measurements do you keep?
Right now we do about 20. We collect a lot of points and then apply rules to detect which body part is which. Consumers don’t need every measurement point, but for the application to help the retailers, I think they do. For example, a tailor is going to need 20-something measures of you. For us to recommend a size to you, we’re going to need a lot more measurements than just bust and hip.
So for instance, if you go to a website like Macy’s and you don’t know what size you should buy for the clothes you’re looking at because you’ve had a very frustrating experience before that when you’ve returned a lot of things. You click a button to authorize the use of your body data and then the algorithm recommends the size you should buy.