Me too (doing some AR stuff)! from Anatoly Zenkov on Vimeo.

Incredible stuff by Anatoly Zenkov.

We are just starting the search to find an extremely talented full-time engineer to join our team in Boulder.

The spot will be an extremely crucial one - we need someone that’s both incredibly talented in our areas of focus, but also able to work alongside the founders at breakneck pace and long hours. You’ll need academic prowess and startup mentality. We’re looking for the best of the best.

We have mind-bendingly exciting products already in the works, but if we’re successful in finding the right person, we expect this person will be instrumental in helping us craft our future products, as well as the future of mobile visual search.

See the Founder-level Computer Vision Engineer posting for more information. And if you think you’ve got what it takes, apply online.

We believe that the future of mobile search is visual - using the camera in your mobile device to to get information about any object or landmark in your vicinity with no typing necessary. Not only will we recognize things via the camera, we’ll be doing it in realtime and overlaying results in augmented reality fashion. We launched our first mobile visual search application, RedLaser, about a month ago.

Today we’re excited to announce that we’ve released RedLaser 2.0, which lets you scan barcodes in realtime. This means you don’t have to take a photo, just hold it over a barcode and we scan it from the video feed. It’s available on the App Store, and we’ve cut the price temporarily to $0.99.

Here is RedLaser 2.0 in action:

Currently the realtime mode only works on UPC barcodes (which is most barcodes in the US), but we’ll be adding EAN (books/European barcodes) support soon. We’ll also be releasing a new version of our SDK that gives developers access to the new mode.

Barcodes are just the beginning, look for more mobile visual search technology from Occipital soon.

The Occipital team is extremely proud to announce, as of this writing, the availability of RedLaser in the US and UK Apple iPhone App Stores for $1.99 USD.

RedLaser has been quietly in development at Occipital since our last release. It’s the first accurate UPC/EAN barcode scanner for the iPhone. While barcode scanning might sound simple, it’s surprisingly challenging given iPhone images — in particular, out-of-focus images.

After scanning a barcode, RedLaser queries Google Product Search, and returns online prices. It also lets you pull up results on Amazon via Safari, and email a product list. There’s also an SDK for developers.

RedLaser is also our first mobile visual search application. Mobile visual search is about visually querying the world in your vicinity. Today, that means you can point your phone at a barcode and we’ll help you find information about a product. In time, we’ll be expanding the possibilities, helping you find information faster and more precisely (oh, and it’s definitely more fun than typing!).

Here’s a short video of RedLaser in action, enabling quick in-store price comparison:

We hope you like using RedLaser!

Today’s pedestrian navigation relies heavily on top-down maps, and hasn’t yet made the leap to 3D point-of-view navigation like many in-car systems have.  A major reason for this is that GPS+compass navigation lacks the precision needed to create a compelling experience on a handheld device.  Several meters and several degrees of error creates an unpleasant jittery experience — even in-car systems occasionally make mistakes about what road you’re on — and the acceptable precision on the road is a lot lower than on foot.

So how do you build an inexpensive navigation system that is sub-meter precise, with degree-accurate heading?

Four months ago, Occipital built a system that is able to achieve this level of precision, by using standard mobile video as an auxiliary position sensor.  After an approximate GPS position is established, video frames are transmitted to a server, where they are compared against a vast database of street-level imagery captured by earthmine.  Each earthmine image is backed by a dense 3D point cloud.  Using all of this information, we are able to estimate the user’s position within a meter, as well as three precise angles of orientation (6 degrees of freedom altogether).

Tomorrow’s pedestrian navigation won’t be top-down; it will be superimposed on the world in front of you.

Similarity transforms come up often in computer vision and gestural interfaces.  The results below are a straightforward application of definitions, but didn’t come up in a quick web search, so I though they may be useful to document here on the blog.

In 3×3 matrix form, a similarity transform parameterized by a scale, rotation, and translation is represented as:

[s*cos(th)  -s*sin(th)   tx ]
[s*sin(th)   s*cos(th)   ty ]
[    0           0       1  ]

Say you have two similarities, T1 and T2, with their parameters as follows:

• scale: s1,s2
• rotation: th1,th2
• translation: (dx1 dy1), (dx2 dy2)

We had a great time this summer as one of the 10 companies that participated in TechStars. I took a lot of photos, and I finally got around to uploading them to the internets. However, these photos are a skewed view of TechStars because I mostly took photos when we weren’t working. So consider the following photos mostly a story of the 5 to 10% of the time we were goofing off.

Our team from the summer. Left to right: Gandhi, Jeff, Vikas, and Shun. This is actually one of Andrew Hyde’s photos.

Soul Patch Concert

Unwinding at the St. Julien after the first demo day:

TravelFli showing off their EntrepreneurWear t-shirts, a counterpoint to VCWear.

Foodzie with Howard Diamond, one of the TechStars mentors.

Rock Band! Andrew Hyde is rocking out on the drums as usual.

Our official company book, it goes everywhere with us.

Andrew Hyde and Jeremy Tanner hanging out at the Bunker.

Jeff and I on stage at Boulder New Tech Meetup.

Hungry entrepreneurs feasting on Foodzie food ferociously. Alliteration anyone?

Tailgating in the parking lot before a Live/ Blues Traveller concert at Red Rocks with Ignighter.

Team Ignighter vs a hybrid Occipital / TravelFli team in a beer pong tournament at our house on the Hill in Boulder. This tournament was where BHARD was born.

Samantha Murphy, the Highway Girl, being amazing on stage at the Laughing Goat.

Our sublet for the house on the Hill ended 2 weeks before the final demo day, and luckily Foodzie let us crash at their place. Gandhi and I slept on the living room floor. Here is Gandhi curled up on the living room floor.

Shun and Jeff slept in the extra bedroom.

Practicing for demo day:

Foodzie’s awesome and adorable dog, Bently, who woke us up every morning in the living room.

The team practicing for demo day:

TravelFli prepping for demo day:

Foodzie with Brett Jackson.

A pizza box with a phrase very similar to “BuyPlayWin“.

Finally it’s Demo Day! Here’s Jeff on the way to the theater.

Team Ignighter looking sharp. Adam, the one in the futuristic coat, is saying “this guy knows what I’m talking about!”

TravelFli dressed appropriately.

Jeff presenting.

And then to unwind! Here we are at the Kitchen Upstairs.

Molly eating an entire ramakin of ketchup (thanks for teaching us that word Austin!).

Austin asking “who wants to go to Houston?”

Nerding it up at the Bunker by playing StarCraft with Devver and BuyPlayWin.

Occipital Founders vs Interns ping pong battle.

David and Jill Cohen invited us all to their beautiful house for an end of TechStars party.

David Cohen, executive director of TechStars, manning the grill. He’s saying “eat more, faster!”

T-Shirts from some of the companies.

Joe “two beer” McVicker from App-X talking to Tom Keller of IntenseDebate.

Ben from Devver saying to his girlfriend, “I’ve got something really important to tell you, my real job is as a backup Britney Spears dancer.”

And here is Ben showing off those dancing skills in the Cohens’ backyard.

And finally, the team at the end of TechStars having a well earned drink. It took me the entire summer to get Shun and Gandhi to drink with us. We miss you guys!

There has been a lot of uneducated speculation about whether the Android G1 could support multi-touch.  I just came across this post by a smart guy who recompiled the synaptics touchscreen driver and got it to track two fingers. Bravo!

No stupid YouTube video like previous attempts at proving this, just some clean console output (trimmed):


x  688, y 3921, z  17, w  1, F 0, 2nd: x    0, y    0, z  17, w  1,
x  696, y 3892, z  21, w  1, F 0, 2nd: x    0, y    0, z  21, w  1,
x  700, y 3887, z  24, w 13, F 0, 2nd: x    0, y    0, z  24, w 13,
x  700, y 3868, z  53, w  7, F 2, 2nd: x 2859, y 1168, z  53, w  7,
x  697, y 3867, z  61, w  6, F 2, 2nd: x 2863, y 1162, z  61, w  6,
x  692, y 3864, z  64, w  7, F 2, 2nd: x 2863, y 1166, z  64, w  7,
x  683, y 3860, z  67, w 10, F 2, 2nd: x 2867, y 1170, z  67, w 10,
x  677, y 3857, z  68, w 11, F 2, 2nd: x 2867, y 1170, z  68, w 11,


We need things to go fast.  Really fast.  So when came across Ryan Block’s old post about the iPhone’s vector floating-point coprocessor, I was encouraged.  But how to access this new coprocessor?  I was excited to find some examples on Matthias Grundmann and Wolfgang Engel’s Google code project, vfpmathlibrary.

They’re just getting started and only have 4×4 matrix operations coded up so far.  Hopefully we can collaborate with these guys to expand the library and do some performance testing on different applications.  More to come on that.

I just came across Miguel de Icaza’s post about recent Mono performance enhancements.  I’m very happy to see all of the innovation in the C# language by Microsoft (see C# 4.0), but the .NET runtime isn’t getting any faster, unless you’re a dynamic language guy.

However, the Mono runtime is getting faster.  They’ve got static compilation, a brand new JIT compiler, and a framework extension so you can easily take advantage of the SIMD (vector) operations that have been available on Intel CPUs for a decade.

A simple way to think about it is that, for instance, you could use the Vector4f class (4 floats) and perform a floating-point vector operation four times faster.  I can’t wait to work with the new Mono.