I guess, what use cases would it make more sense to use an anonymous counting system rather that one that also acts as a security system (where we can identify suspects...) Just finished watching the video, I still feel like in most of those cases (maybe minus the conference room) that it would be nice to double up as security...
@stewartjarod There are two reasons we've come across - perception and functional.
Perception -- when you're anonymous, you can go places a surveillance camera can't. We've heard from stadiums about bathrooms, publicly traded companies that don't want third-party vendors storing video off-premise, churches (turns out people count is how most dioceses dole out capital, college campuses that can't have their students "surveilled," coffee shops and other SMBs who have an aversion to "tracking customers" as a way of gathering relevant data.
When a location uses a camera for security and play-back, it's justifiable. When you use the same camera to gather data about foot traffic, or do facial detection, or demographic estimation, it starts to cause privacy issues and can hurt patronage.
Functional -- there are a lot of surveillance cameras from a lot of different providers already deployed. Getting them to work properly with one another is a significant challenge. There are computational problems like "hand-off" - when a person moves from one camera's field of view to another. This blind gap in-between those Fields of View needs to be algorithmically accounted for in order to not lose track of the person.
You'd also need a lot of cameras installed to cover a large area. There are bandwidth concerns - image processing would have to happen on-device in order to not hog a network. Lastly from a computer vision stand-point, separating all the different kinds of backgrounds from moving objects and correctly categorizing them is not a small task. There are a lot of cameras that do this but there's still basic research going on in right now in university trying to solve this problem.
In some cases, we've actually found customers who would like to use Density in conjunction with a closed-circuit security camera system (like tailgating in secure offices)
A sensor dedicated to deriving relevant information without compromising consumer privacy has been a big selling point for our customers. It's also easier to explain.
Some other uses - https://www.density.io/uses/
Hope this helps!
@abe_storey We cannot determine male v. female. But by design. Anonymity has been an important selling point for many of our customers. Definitely let me know if you'd like to get one andrew@densirty.io
@michalnaka 100%. We've heard from bus systems, subways, airports, and trains. Long term, we'd love the data to be available to the organizations who deploy as well as their travelers. I think there are some really interesting opportunities in open data.
@andrewfarah absolutely - my company works with transit & mobility providers and users around the world. Would love to keep in touch about this. working with a few cities right now on open sourcing occupancy data.
@andrewfarah indeed we did. As your very first pilot customer, I'm proud to have been witness to such great progress. You've come a long way my friend and it's impressive!
@adamroot Appreciate the note. That's high praise. Although... it is summer and tech gets a little sleepy around August :). Why did you find it interesting? Data, business model, device, something else? Would love to hear about your background.
Product Hunt,
I'm Density's CEO, Andrew.
The product in today's post has been about two years in the making. I remember @rrhoover hunted our cobbled together version 1.0 back in 2014.
In our first mass market sensor, we've combined a powerful people counter, a modern API, a dedication to privacy, and our sensor-as-a-service business model. The hardware is free. Users pay a monthly fee for access to the data. Prices start at $45 / sensor / month.
Density uses depth technology, computer vision, and an onboard quad-core processor to anonymously measure and manage entrances and exits through a door. The tech is intelligent enough to handle complex human behavior -- groups, collisions, bi-directional movement, lines and lingering.
Some gifs on what the sensor sees -
Group detection - https://giphy.com/gifs/l0HlKvlJC...
Anonymous by design - https://giphy.com/gifs/l0HlSnYdQ...
Ignoring a door - https://giphy.com/gifs/3o6ZtnFWL...
We'd love your thoughts and critique on the physical product, business model, customer types, etc.
With much ‹3 for the PH crew,
Andrew & the Density Team
@andrewfarah Funnily enough, I was having drinks with the IT director of a mega church a couple of nights ago. They're beginning to delve into data science and analytics to manage their five congregations of 40,000 people or so. I'm uncomfortable with religion in general, but it was fascinating to think about church attendance in a startup-y sort of growth model. Right now I believe they manually count the empty seats in the church to estimate attendance. We were talking about the idea of using cell phone mac addresses to measure retention, running A/B tests across their 5 locations etc, but I bet something like this would be really useful to them too.
You can see more about Halo Neuroscience here: www.haloneuro.com. We just opened pre-sales on our product yesterday!
@andrewfarah@rrhoover This is a simple and useful product with a clean design. And the data could be very useful in an emergency, besides you could add some triggers to platforms like IFTT or Zappier, so people could get notifications when their favorite place has enough people or even trigger IoT devices... sorry if I dream a lot but I see a lot of uses in a simple and clever device like yours.
I have to say, I follow your product since the early days (also remember when you guys shared the logo research you made, which I've since used and integrated in my logo design process). Beautiful new product, excellent design and overall great team, congrats on the launch!
@widawskij Ah! The Medium article we wrote ages ago! That's awesome that you remember that. Thanks for the support along the way.
What do you do? Any way we can be helpful?
Hits the sweet spot on each tech trend as of today.
1. Pay for service, not hardware.
2. Use AI, not imperative code.
3. Realtime IoT API and not download-data-at-end-of-day.
Looks very cool, well designed and nice use cases! Curious to hear why you chose for a monthly subscription model instead of a one-time fee. Can think of pro's and cons for both sides and curious why you ended up going for the subscription model.
@freekgille There are definitely pros and cons to a sub model. The downside is that a customer doesn't get to "own" the hardware, has a recurring fee, etc. The upside is that we can learn from the hardware, swap the hardware when tech gets better, focus on selling the data as opposed to the cost-of-goods-sold + some margin.
What do you see as pros and cons?
@andrewfarah Would love to highlight this -- and show a use case, get you some customers/devs -- at API World (http://apiworld.co) in September (12-14) in San Jose. Send me a note.
@andrewfarah What would the likelihood be of a future version be that tracks, but doesn't capture, identity? I run a movie theater and I could come up with some amazing insights if I knew that anonymous person A went to movies alpha, delta, and gamma, and person B went to movies beta and lambda. I don't need to invade their privacy in order to learn from their collective behavior. :)
Very cool! I'm a huge fan of hardware being used to provide great value in industrial context. The especially interesting thing about this is that it's valuable to both the business itself (understanding and optimizing for traffic patterns) and the patrons (finally I'll know when the gym is too full to bother)!
I'm excited to see where this goes.
Dropbox DocSend