Crowdsourced content is only as good as the ability to curate it. Mapillary does exactly that by using computer vision to understand the relationships between two images and stitching them together automatically. Using GPS information, it is building the largest user-generated photographic map of the world. We caught up with co-founder Jan Erik Solem to learn more before he takes the stage at the LDV Vision Summit on June 4th, in New York.
First, here is how it works :
What is Mapillary and why did you create it ?
It is a service for crowdsourcing street level photos for maps. With simple clients collecting the photos, we use computer vision behind the scenes to make sense of and connect the photos into something meaningful. The system is open for many data sources as input and anyone can build viewers for interacting with the photos.
It is an idea I have had for a long time and finally the timing was right.
Isn’t it competing with the already very popular Google Street view ?
We see it more as a supplement right now but it could grow into a
competing service over time. The main problem we are trying to address
it the lack of coverage and update frequency in some parts of the
world. Some areas are just not cost-effective to map with cars.
Crowdsourcing doesn’t have the same limitations. As a contributor, you
can take as much detail as you like, as often you like.
Who is your typical users ?
Right now, very much an early adopter crowd. People interested in
mapping, some OpenStreetMap contributors, people passionate about
their home town or neighborhood. Some share from their vacation or
from hikes and bike rides. Some obviously try to map every single
corner of their area.
How does Mapillary plan to generate revenue ?
Commercial API plans and private repositories. The latter means that
you can have a private part of Mapillary only visible to your
organization or team.
Can you give us some insight on numbers : how many users ? How many
images uploaded ? How fast is the company growing ?
Today, we have close to 700k photos and 10000 miles mapped by about
2000 users. As a company we are steady at 4 people now. Might grow a
little over the next few months.
Tell us more about your involvement with the LVD Vision Summit . Why
did you decide to collaborate on the event ?
I have known Evan for a while and we have collaborated an co-invested
in the past. When he brought up the idea of the Summit I thought it
was a great idea. I’m looking forward to a different kind of
What do you expect to gain from it ? In other words, what would be a
key sign of success.
The mix of investors, startups, bigger companies, researchers, all
within a tightly defined field of “imaging” is very interesting. I’m
hoping to make new connections. Good signs of success for the event in
general would be if new connections lead to concrete action. Startups
getting funded. Researchers getting jobs or starting companies. Things
Your LinkedIn profile says you like to work on large-scale computer
vision problems. Are they many?
Many problems in computer vision are turning into large-scale problems
today. Some examples: Image search and retrieval was an early one that
is still growing in size. 3D reconstruction is now done on city level.
Organizing a professional photographer’s library is hundreds of
thousand high-resolution photos these days. Making sense of life
logging photos will be insanely big in terms of photos. At Mapillary
we are analyzing and finding connections between hundreds of thousands
of photos. And so on…
With Polar Rose, the company you sold to Apple, you were at the
forefront of image Search. Where do you see the field now ?
We were focussed on faces and that was new at one time. Today, things
are moving to more general recognition and tagging of image content.
Search is just one aspect that benefits of that technology maturing.
What would you like to add to Mapillary today that can’t be done
because of technology restrictions?
A smart camera with great optics and all-day battery that
automatically understands when photos need to be taken. Just wear it
or attach to cars and the rest is automatic.
You can catch Jan Erik Solem in person at the LDV Vision Summit on June 4th in New York. Tickets, as well as more information, are available at the event website here. You can also download the free apps for Android and iOS.
Author: Paul Melcher
Paul Melcher is the founder of Kaptur and Managing Director of Melcher System, a consultancy for visual technology firms. He is an entrepreneur, advisor, and consultant with a rich background in visual tech, content licensing, business strategy, and technology with more than 20 years experience in developing world-renowned photo-based companies with already two successful exits.