IQ Engines

Photo Hack Day 4 Recap

iqengines-mark — Thu, 11 Apr 2013 17:08:00 -0700

We had a great time participating in Photo Hack Day last weekend at Facebook headquarters. Aviary ran another outstanding event, attracting hundreds of developers and 20 APIs. Out of the 63 hacks presented on Sunday afternoon, 8 teams used IQ Engines. This year in addition to promoting our API as a sponsor, 3 of our developers decided to build their own hack.

We were thrilled by the enthusiasm and creativity of all the teams who used our API. Our choice for best use of the API was Photo Journal, a web service that automatically tags photos of beach, mountain, snow, sunset and other scenes; then creates personalized albums based on a common theme (i.e. beach) and adjusts the design to complement that theme.

Thanks to all the hacks who used our API and kudos to IQ Engines own team Billie, who took home prizes from Chute and imgur for their nifty bill splitting app. For more, check out this article on the event as well as the PHD4 photo wall.

Scene Recognition Hack Ideas for Photo Hack Day 4

iqengines-mark — Fri, 05 Apr 2013 11:00:55 -0700

This weekend at Photo Hack Day 4, we’re offering exclusive early access to our Scene Recognition API. Use it to automatically tag beach, mountain, night, snow and sunset scenes in your hacks. Here’s a few ideas to inspire you.

Stock photo finder: search for stock photos by scene
Smart albums: make photos on your phone searchable with auto tags
Auto filter: apply filter/edits automatically based on scene labels
Live scene browser: view photos that are being uploaded in real-time by scene
Collage maker: create a collage of your favorite sunsets, edit and print

OCR Hack Ideas for Photo Hack Day 4

iqengines-mark — Thu, 04 Apr 2013 16:15:00 -0700

Going to Photo Hack Day 4? Get exclusive early access to our OCR API this weekend only. Use it to recognize text in photos, ads, receipts, coupons and more. Here’s a few ideas for your hacks:

Bill splitter: scan a receipt to divvy up the bill or log an expense
Photo finder: index text in online photos to provide a new way to search albums
Visual check-in: take photo of sign, menu, logo or storefront to check-in with Foursquare
Auto dialer: take a photo of a phone number to call it
Sign translator: photograph a sign or menu to get a translation
Connect: take a picture of a business card to add to your contacts or social media

Join IQ Engines for Photo Hack Day 4

iqengines-mark — Thu, 28 Mar 2013 10:41:20 -0700

Join IQ Engines for Photo Hack Day 4 on April 6-7 at Facebook HQ in Menlo Park.

Meet our rockin’ developer team
Be first to try our OCR and scene recognition APIs
Get help building a badass photo hack
Compete for over $15k in cash and prizes
Enjoy a full weekend of photo-centric awesomeness

Browse featured APIs and register for free at: http://www.photohackday.org/

We look forward to seeing you there!

SmartAlbum Adds Face Detection and Recognition

iqengines-mark — Wed, 20 Mar 2013 10:21:36 -0700

We released an update to SmartAlbum this week that includes some great new features:

Face detection: finds all the people in your photos.
Face recognition: organizes your photos by person.
Group shots: finds group photos.
Portraits: finds close-ups of a single person.
Bike detector: finds photos with bikes.
Car detector: finds photos with cars.
Easier Dropbox uploads: drag-and-drop photos from your desktop to a designated folder in your Dropbox account.

Go to http://www.iqengines.com/smartalbum and click on “Demo” to see for yourself. You can use IQ Engines’ photos or try it with your own photos using Dropbox.

New Year, New Website

iqengines-mark — Thu, 21 Feb 2013 11:29:00 -0800

The new IQ Engines website reflects our focus on SmartAlbum for photo organization and SmartCamera for mobile commerce. Both products are based on IQ Engines’ powerful image recognition platform.

Check it out at www.iqengines.com and try the demos to see how:

SmartAlbum makes it easier for people to browse, search and share their photos.
SmartCamera makes it convenient for people to “point, snap and find” products using their mobile phones.

Join us Wed Night at the Parisoma Retail Mixer in SF

iqengines-mark — Mon, 10 Dec 2012 15:08:00 -0800

IQ Engines will demo our mobile visual search solution for retailers this Wed, Dec 12th at the Parisoma Retail Mixer in San Francisco. Find out more and sign up to attend the event here: http://bit.ly/T2nkJf

Tesco does Mobile Visual Search

iqengines-mark — Wed, 05 Dec 2012 18:50:00 -0800

IQ Engines now powers mobile visual search for UK-based Tesco, the world’s third largest retailer. The Tesco mobile app makes it easy to shop for groceries on the go. Just take a picture or scan the product’s barcode to get more information or add items to your shopping list.

IQ Engines Announces Smart Album API

iqengines-mark — Tue, 13 Nov 2012 11:00:00 -0800

IQ Engines is releasing a new Smart Album API that uses image recognition to make it easier for people to browse, search and share their photos. The beta release recognizes six types of scene (beach, food, mountain, night, snow, sunset), landmarks (Golden Gate Bridge as an example in our demo) and the presence of text, people and faces in photos. More scenes and classes (bikes, cars, dogs, etc.) will be added in subsequent releases - let us know which ones you’d most like to see!

The Smart Album API is for anyone whose business revolves around storing, managing and retrieving images. Our API makes it easy to integrate with your photo website, desktop photo management software or mobile application. It includes innovative new features such as photo stacking, auto tagging and face detection that take the drudgery out of managing online photo albums.

Try our online demo or contact us at support@iqengines.com to find out how Smart Albums from IQ Engines can help you:

Improve your photo management website, software or app
Auto-tag large photo collections so they’re sortable and searchable
Offer a better way for people to browse and search their photos
Reduce the time it takes people to build photo albums so they’re more likely to share
Increase personalization and deliver more relevant ads based on photo content

Photo Stacking - reduce clutter
Anyone who has ever taken the same photo ten times to try and capture the perfect image will love “stacks”. With photo stacking, similar and duplicate photos are automatically grouped together. This reduces clutter and makes it easier for people to scroll through their albums.

Auto Tagging - improve browsing and photo search
Tagging photos manually so you can find them later is time consuming. We use advanced computer vision techniques to identify and label each photo, automatically. Once the photos are labeled, it’s easy to sort, search and browse based on what’s in the photos. With auto tagging, your users can search by scene, landmark and even text without entering any data themselves.

Face Detection - find photos with people
Our smart album software can also detect people and faces. This makes it easy for users to find portraits and group shots among all of their other photos. In future releases, we’ll add facial recognition to allow users to search photos by name.

NIH Funds Mobile Search for the Visually Impaired

iqengines-mark — Fri, 26 Oct 2012 16:39:00 -0700

Grants from the National Institutes of Health (NIH) provided an important early source of funding for IQ Engines. These grants helped fund development of image recognition technology that can help the blind identify objects they encounter, track the items they own, or navigate complex new environments. Several companies have developed mobile phone applications and dedicated devices for the visually impaired using this technology.

The IDEAL Currency Identifier is an Android app from Apps4Android that enables people who are blind or visually impaired to denominate U.S. currency. It will be expanded to identify the Euro, Swedish Krona, Swiss Franc, Australian Dollar, and Canadian Dollar.

VizWiz is an iPhone app that lets blind users recruit remote sighted workers to help them with their problems in near real-time. Users take a picture with their phone, speak a question, and then receive multiple spoken answers from a network of crowdsourced workers.

i.d. mate Quest from En-Vision America, is a portable “all-in-one” talking barcode scanner that aids visually impaired individuals with the identification of items via the product’s barcode or UPC. In addition, an on-board camera provides features such as currency identification and object recognition enabled by IQ Engines.

For more about how NIH grants helped IQ Engines develop mobile search technologies for the visually impaired, read the SBIR and STTR Success Story for IQ Engines.

VocaBoca Visual Translator, powered by IQ Engines, wins International Create Challenge 2012

iqengines-mark — Thu, 25 Oct 2012 16:22:00 -0700

VocaBoca won the International Create Challenge for a language learning app that uses IQ Engines’ image recognition technology. VocaBoca lets students take a picture of any object with their smartphone’s camera to get a translation, as well as dictionary and encyclopedia definitions, in less than 30 seconds. The app lets students compile custom vocabulary lists using automatic language detection, speech recognition, text-to-speech and image recognition technology. For more information read the press release or watch the VocaBoca Pitch Video.

IQ Engines is hiring - see our open positions now

iqengines-gerry — Thu, 18 Oct 2012 16:24:00 -0700

We are hiring! If you are a kick-ass software programmer or computer vision researcher, come join us and help build the world’s best image intelligence platform. View our open job positions.

New Android App Uses Image Recognition to Help the Blind Identify U.S. Currency

iqengines-mark — Tue, 09 Oct 2012 18:01:00 -0700

The IDEAL Currency Identifier is a new Android app that uses image recognition technology from IQ Engines to help the blind identify U.S. currency. The app was developed by Apps4Android, Inc., a subsidiary of IDEAL Group to fulfill a government mandate to provide increased access to U.S. currency for people who are blind or visually impaired.

Watch this video to see how the app lets blind users scan a bill with their phone’s camera to hear the bill’s denomination read aloud.

For more information, read the U.S. Department of Education press release which credits IQ Engines for the advanced image recognition technology that made the app possible.

Image Recognition with Multiple Results & Multiple Object Recognition

iqengines-mark — Fri, 28 Sep 2012 10:31:00 -0700

Check out this video for a look at two new features in our reference app for visual search - image recognition with multiple results and multiple object recognition: http://vimeo.com/50340681

Multiple Results

Now when you scan an object, if our image recognition algorithm finds more than one visual match, oMoby provides multiple results. The top result is listed first, followed by other close matches in order by strength of match. The multiple results feature improves the performance of our image recognition service by reducing false positives and increasing the likelihood users will get relevant results from their visual search.

For example, in the case of wine labels with very small distinctions between a merlot and chardonnay from the same vineyard, multiple results will provide a list of all matching labels.

Multiple Object Recognition

Now oMoby also detects and recognizes multiple objects in the camera’s viewfinder, simultaneously, with a single visual scan. This means users can run a visual search on a scene with many objects and get results for all of them.

Download the latest version of oMoby today to try these features for yourself. Let us know if you’d like to add image recognition with multiple results or multiple object recognition to your app; they’ll be available soon as part of the VisionIQ iOS SDK.

IQ Engines Powers Winner of Best Mobile App at Facebook World Hack Taipei

iqengines-mark — Wed, 26 Sep 2012 15:56:00 -0700

A team using IQ Engines’ image recognition API won best mobile app at the Facebook Developer World Hack in Taipei earlier this month. The social travel app, “Before Sunrise, Before Sunset”, lets people plan their trip and share the experience with friends using information from their social network. Before the trip, travelers create an interactive t-shirt with images of people and places they want to share. Then, people they meet on their travels can scan the t-shirt images to link to the traveler’s Facebook profile, get more info on places or even check-in (using image recognition as the trigger).

The app uses on-device image recognition from IQ Engines so it works even when there’s no internet connection by saving the data people and places it recognizes for later. The team at Linkwish tells us they’re planning to bring the app to market later this year and replace the t-shirt with printed scannable cards. Here’s a video of the team’s presentation at World Hack: http://bit.ly/QDxHAT (in Chinese).

About Linkwish

Linkwish is an app development company in Taiwan, whose goal is to help people connect more in the real world. Linkwish developed two apps MOS Order and MOS Wish for the second largest restaurant chain in Taiwan, MOS BURGER. “Before Sunrise, Before Sunset”, winner of best mobile app at Facebook Developer World Hack in Taipei, was built in six hours by Bradley Lin, Bob Wei, Ellen Chen, Jinyao Lin, William Hsu, and Ju Chun Ko.

IQ Engines Demos Smart Photo Albums and Produce Recognition at Bay Area Vision Meeting

iqengines-mark — Wed, 19 Sep 2012 14:52:00 -0700

Last week IQ Engines attended the Bay Area Vision Meeting at Stanford University. Turnout was great with almost 400 computer vision enthusiasts from industry and academia in attendance. The focus of the event was on “Large Scale and Fine Grained Recognition” and included speakers, posters and demos on related topics.

We showed off our latest scene recognition and class recognition technologies with demos of smart photo albums and produce recognition, which impressed the tech savvy crowd.

Smart Photo Albums - scene recognition technology
Our smart photo album concept illustrates how scene recognition technology can be used to make it easier for people to organize, browse and search their photos. By running our scene recognition algorithm on hundreds of photos we can automatically label each image as either outdoor, close-up, food, indoor, night, sunset or document. Once the photos are labeled, users can easily sort, search or browse their photos by scene type.

Stacks are another feature of smart photo albums that allow users to reduce clutter by grouping similar photos together. In the example below we show an album with stacked photos (the number indicates how many photos are in each stack) and another album with four similar photos unstacked.

Produce Recognition - class recognition technology
To show how our class recognition technology handles three dimensional objects we demoed an iPhone app that recognizes different types of produce such as avocados, grapes and melons. Whereas the instance matching techniques available in our commercial API work best on flat media such as packaged products, class recognition will extend our capabilities to recognizing 3D objects like bikes, cars, people, and of course, produce.

For more information on the Bay Area Vision Meeting go to: http://vision.stanford.edu/bavm2012/

Here’s a peek at one of the new features we’ll be...

iqengines-mark — Mon, 10 Sep 2012 10:41:00 -0700

Here’s a peek at one of the new features we’ll be adding to the VisionIQ image recognition API: “streaming visual search.” This feature makes it easy for users to scan multiple objects, simply by pointing their camera - no need to press a button. Streaming visual search automatically looks for objects to be recognized and continuously streams the queries to IQ Engines’ servers. Streaming visual search provides a similar user experience to a feature that is already available in our API for on-device scans: “continuous visual scanning.” However, streaming visual search has the advantage of scaling to recognize millions of objects because it uses the VisionIQ API for cloud-based image recognition.

Monitor VisionIQ operations with our new service status board

iqengines-huy — Fri, 24 Aug 2012 16:28:00 -0700

We’ve worked hard to make sure that VisionIQ is the fastest, most scalable, and most stable image intelligence platform in the world. But like all systems that are rapidly evolving, ours is not immune from occasional abnormalities or downtime.

To make it easier for our customers to monitor VisionIQ performance and to keep you informed on the rare occasions that errors do occur, we now provide a service status board for the API at status.iqengines.com. Bookmark this page and refer to it for up-to-date info on our platform status.

Green (operational), Yellow (degraded performance) and Red (non-operational) indicate the status for each major component within our system
Event log shows the status history
If you see yellow or red on any component, we know about it and are working to resolve the issue as quickly as possible. We’ll update the event log as soon as we get more information, so you can refresh the page to get periodic updates on the situation.

Our commitment to quality

As a strategy to minimize downtime and reduce the impact of occasional hiccups on the API Service, we as an engineering team ask ourselves two key questions when building new systems or improving existing ones:

How do we minimize the amount of time it takes to discover abnormalities or errors?
How do we minimize the amount of time it takes to recover from errors?

The answers to these questions vary with each project, but generally we aim to make sure our apps and services are:

monitored
covered through extensive unit testing
benchmarked
repeatedly deployable with little-to-no human intervention
have built in fail-over mechanisms

It’s our belief that building software with this approach leads to higher quality systems which translates directly to faster, more responsive, and reliable service for our customers.

Are you an engineer interested in devops and building robust, scalable, vision systems? We’re looking for talented people to join our team. Check out our job boards for more info.

Check out the latest image recognition technology from IQ Engines Labs

iqengines-mark — Tue, 21 Aug 2012 10:38:00 -0700

The new “Labs” page on the IQ Engines website provides a rundown of some of the cool features we’re working on to make VisionIQ image recognition technology faster, better and cheaper. Read about scene recognition, OCR, class recognition, visual similarity and streaming search; and be sure to check out the demos of each technology in action. These capabilities are currently in development, but stay tuned. We’ll let you know as soon as they’re ready for primetime.

https://www.iqengines.com/labs/

Image Recognition Hacks at Photo Hack Day 3

iqengines-mark — Thu, 19 Jul 2012 14:20:00 -0700

IQ Engines was one of 21 APIs participating in Photo Hack Day 3 this weekend in San Francisco. This was the first west coast trip for the popular New York City event and turnout did not disappoint. Over 300 developers showed up at Dropbox’s cavernous San Francisco headquarters, creating 70 hacks in just 24 hours. Aviary, which makes a simple photo editor that can be embedded in any app or website, ran an outstanding event and we look forward to more PHDs in our future.

For IQ Engines, the highlight was hearing all the cool ideas developers had for using image recognition in their hacks - from crowd counters to auto tagging products in Facebook, to cat removers and baby detectors, and much more. Here’s five interesting hacks using IQ Engines that teams came up with during the event. While not everyone succeeded in carrying them off, we loved the inspiration and creative hacking.

more.io built a mobile phone app to help people discover items they may want to buy. When you see a product you like, take a picture with more.io. This triggers a request to IQ Engines’ image recognition engine, which returns a tag (i.e. “orange dress”). More.io searches Etsy for orange dress and returns a list of similar products.

Tapsee is a mobile phone app that identifies currency and everyday objects for the blind and visually impaired. It combines iPhone VoiceOver accessibility features and image recognition. Users open the app then double tap the screen to take a picture of medicine, packaged food or currency. IQ Engines recognizes the object and Siri speaks it aloud.

Photo Assassins is a mashup of two popular games: Assassins and Photo Scavenger Hunt. Players are assigned a set of photos, which are the designated targets. Each player tries to guard their own target and assassinate other players’ targets. To assassinate a target you must take an unobstructed photo. If a bodyguard shows up in the picture, IQ Engines will not verify the photo and the target remains alive.

Sudoku Go aims to let people take their sudoku puzzle from the newspaper to their mobile phone and even a Facebook app. Players scan the game in progress with their phone and Sudoku Go converts it to a playable mobile game. Later, if they’re not finished, players can open the Sudoku Go app online to continue playing from within Facebook.

Augmented Popup lets people create and share augmented reality popup books. Just point your camera phone at a page in the book. Then add photo popups that appear layered over the book page. Photos can be added from photo feeds by scanning a real-world object to run a visual search on 500px or Flickr.

For more on Photo Hack Day 3 read Aviary’s recap http://bit.ly/NBwTwq and Filepicker’s overview of interesting hacks http://bit.ly/OEkLb9.