kooaba has been mentioned in several Gartner report before, but being featured as a cool vendor is … well, simply cool.

The iZeeum app interaction flow

Using visual recognition makes selection of the right track extremely easy, furthermore using your smartphone to play audio guide content or read about exhibition items is a much smoother experience that using classic audio guides.

Very fitting iZeeum launched their app with the photo exhibition “Doug Menuez: Fearless Genius. The Digital Revolution in Silicon Valley” in the Coalmine gallery in Winterthur, Switzerland. The exhibition is open until June 2013 and the iZeeum app is a free download in the Apple App Store.

In the case of Google Glass we are beyond the state of rumors already. Specs seem pretty clear and a release is announced toward the end of this year. Bloggers have even had a chance to give it a spin.

A Google employee wearing Google Glass. Credits: Mathew Sumner/Getty Images

The scope of the iWatch project is much less clear. If it even exists. However, considering the secrecy of Apple around its product launches this is not surprising.

So why these rumors, and why now?

Companion devices for your smartphone

If anyone attended last weeks Mobile World Congress in Barcelona, she will have noticed that smartphone technology has reached at least some local maximum. Devices made by newcomers such as Huawei or ZTE reach the same capabilities like their Apple and Samsung counterparts. The main novelties are bigger displays and incremental improvements. The only mandatory feature still missing is NFC. This industry needs some real news!

Beyond that, focus shifts from hardware to services such as information lookup, assistants, maps etc. Services include Google Now, Siri etc. Simplifying access to those is paramount for future success.

Finally, the smartphone craze with status updates, tweeting etc. has reached a new level which clearly has it’s impacts on society through people pulling out their smartphones at any possible opportunity.

Devices such as Glass or iWatch could solve these three problems. New Hardware, quick access to relevant information and services, and all this in a (stylish?) and unobtrusive way.

So the key insight here is that both devices fulfill the same function. And both devices (most probably) run as companions with your smartphone, which acts as the “brain” and “gateway” to connect to the digital online service world. The only difference is the form factor. And with the form factor there will be different features.

Some core features like a screen, which displays current information, and voice input will be shared. But what about other sensors?

Form factor and implications for visual recognition

Glass has an integrated camera. Saying “OK Glass, take a picture” allows for taking snapshots without even having to open a camera app on a phone. Already at CVPR 2012 Sebastian Thrun (of Google’s equally ambitious Self Driving cars project) reported about Glass, how it lead to another increase of pictures taken and shared (as if 300 million photos uploaded on facebook per day weren’t enough already).

This means, that the digital companion hardware devices could increase demand for visual recognition solutions further. If they include a camera. Admittedly in the case of the iWatch the form factor of a wristwatch makes this hard to imagine. But you never know what the guys in Cupertino come up with. Let’s wait and see.

The Shortcut app of our sister company Shortcut Media is just one example of such an app. A similar use case are interactive print catalogues, that link the physical catalogue with mobile shopping options.

In contrast to QR codes, if there is a publication with many pages, making the whole issue interactive just becomes a matter of adding pages to the recognition system, and adding additional content to the pages.

But how to you get the pages into the system really quickly? We wrote a little Ruby script for you, which does just that in a few lines of code. What it does:

• split a PDF file into pages
• convert each page to an image
• upload it to kooaba
• store thumbnails and the original PDF of the page to s3

The code can be found on github. We are looking forward to many new interactive print apps.

And btw, if you need a mobile app, check out our iOS tutorial from last week.

The new version of the sample code watches out for a special key (“redirect_url”) in the item meta-data. If you populate the data of this key with a URL in your meta-data, upon recognition of the item this URL is shown directly. You can also switch between a debug and a demo mode. This way you can have an image recognition iOS app ready for demoing to your boss within 10 minutes.

We made a little screencast of the complete workflow. The iOS part starts at minute 3:30.

Well without further ado, here it is:
https://github.com/kooaba/kooaba-api-v4-samplecode/tree/master/iOS

This should make integration in your app really easy. The sample code comes with an associated little app project for Xcode. This way you should have a simple app up and running in a minute.

The code will be continuously updated from now on with additional features. Stay tuned. (e.g. simply watch the repository on github)

For many (mobile) applications knowing about the context a user made a query is essential. Context can include for example the geographic location, a user id, or any other data you consider relevant. This context can be used for serving contextual (e.g. personalised) results, statistics, or just debugging purposes.

The new feature allows you to transmit such user data with your queries. (User here refers to you, the API user). You are free to transmit any kind of data that is useful to you, by adding it to the new user_data field of the query API in JSON format. The data is stored and is then available in the web UI.

We also parse this data for certain parameters that are known to our system. At this time our system reacts to the geographic location (latitude and longitude are the keywords to use here) and to a user identifier (with the key user_id). The idea of the latter is to allow for tracking and grouping queries by users or devices. User id can be any kind of identifier string, for instance a device uuid or an e-mail-address.

These parsed fields are displayed separately in the query views of the web interface.

Here’s an example query executed with curl, which sets the user data to a combination of known and unknown keys:


curl -i -X POST \
-F image=@image.jpg \
-F user_data='{"user_id": "1234", "device_type": "iPhone 5, iOS 6", "longitude": 47.3690, "latitude": 8.5380}' \
-H 'Authorization: Token ' \
-H 'Accept: application/json' \
'https://query-api.kooaba.ch/v4/query'


The query shows up in the backend interface like this:

The query as it is visible in the web UI. Note the extracted fields for lat/lon and user_id

The new version of the API runs on a completely separate system, and has been completely restructured. Why this major new release you may wonder? Over the years we have collected significant experience in building visual search applications, and we collected a lot of input from you, our users. We wanted an API that’s the best of its class, that’s simple to use, and comes with increased scalability. Let us tell you what we came up with.

The new APIs

Here are some of the highlights of the new API.

• the new API’s are simpler to use and more flexible: metadata specified in JSON allow you to store and return any kind of meta data
• the new API’s are faster. We removed a number of processing steps at query side, and made some parts of the calls asynchronous. This way, queries reach recognition servers as fast as possible.
• The API’s are more scalable, in particular on the upload side we can handle more throughput
• Authentication is easier and more flexible: we offer now two kinds of authentication. Besides our signature based authentication we also allow simple token authentication via HTTPs. This makes integration easier, especially on upload side.
• API key management: you can create and manage multiple API keys. There are separate keys for query and upload, and you can create fine-grained permissions for each key. This way you can for instance have several applications with different keys etc.
• The old version of our API allows for uploading data to “groups”. On request we would create multiple groups for you. Since this was pretty popular, we extended this concept. You can still organize your data in what we now call “buckets” and you can now create as many buckets as you want. Permissions for accessing buckets can be set on API key level. So you don’t have to pass any parameters about which buckets are searched at query time. You can change these parameters anytime in the backend UI, without touching your (mobile) app.

This is what’s new with the API. But that’s not all. We also have a brand new backend UI waiting for you!

The new management UI and API documentation

The new management user interface has been simplified drastically. An overview is given below: you start with an overview of your recent account activity on a dashboard. You can browse your queries and manage reference data. Finally, you can configure your API keys as described above. The screenshot below gives an overview:

The user interface is complemented with much more extensive documentation, re-written from scratch. There is also more extensive sample code.

Documentation and sample code can be found here: http://docs.kooaba.com.

(The only major thing missing is currently sample code for iOS: this will be updated soon.)

Migrating to the new API from an older version

Since the API v4 runs parallel to the API v1 on a new system, and the API format changed, you need to sign-up again and reimport your reference data, or update your apps. We are happy to assist you with this process.

You can sign-up for the new API here: http://platform.kooaba.com

API v1 will be kept running for the time being. However, we recommend switching to API v4 at your convenience to benefit from the new features and performance. Furthermore, in the coming months, we expect to add additional amazing new features and algorithms to the new image recognition platform v4.

Should you have any questions get in touch with us via support at kooaba dot com.

For us the year could not have started better. Just end of last year two of our customers told us about some stats that made them really happy – and what makes our customers happy, makes us happy.

Our friends over at Vanija created a really nice Christmas campaign app for Switzerland’s largest retailer Migros. The idea they had was to create a kind of “treasure hunt” app. By snapping pictures of any Christmas related marketing materials (Flyers, in-store ads, products, …) you could collect points and win in weekly sweepstakes. The app was a huge success: ranked in the top ranks of the Swiss App store throughout the Christmas time, and in terms of requests going well beyond the limits of our default pricing plans. Below are some screenshots from that app to give you a better idea what it is all about.

The other good news are from Vivino one of our favorite kooaba-powered apps. Christmas and New Year’s are the best time to drink some great wines and learn more about them using kooaba’s image recognition in the Vivino app. During holidays Vivino reached more than 150k scans per day! Beyond the sheer number of request the engagement that develops around the (photos of) wines is really amazing, as shown in the screenshot below.

This kind of news gives us an extra boost of motivation! So can it get any better? Of course. We want more happy customers, so we have been working really hard on new versions of our platform. Stay tuned, there will be news as early as next week.

Of course, now we want more. The same in other countries, and an even higher percentage. That would make a nice goal for 2013. We’ll keep you posted.