Imagga @ Machine Learning Event at SAP Labs Bulgaria

Georgi & Georgi presenting at SAP Labs Bulgaria

Machine learning is getting lots of attention lately. It’s amazing that some 200 people showed up at Hack Bulgaria event and stayed almost 3 hours to learn more about machine learning. Not to mention it was friday and the venue was not in the center of the city! It was a clear indication for us that lots of developers are getting curious about machine learning (ML) and that’s totally cool for companies like ours.

This is a short overview of our not so tech presentation about machine learning for images.  Some of the other lecturers have covered different aspects of ML and it’s application in various cases and industries.

From the moment of their invention the convolution networks were great for tasks as face detection and handwriting. Thanks to the advance of the GPU technology and the extended base of image data, the convolutional networks demonstrated far better results for complicated tasks such as visual classification of objects in images.

There are some specifics when it comes to image recognition using machine learning. Images are a matrix of pixels (raster data) and that’s why recognition is sensitive to lighting, contrast, saturation, blur, noise, geometric transformations (scaling, translation, rotation) and occlusion.

Machine Learning Meetup Sofia

Conventional image recognition methods struggle to find the optimal set of filters (convolutions) to apply for each specific use-case.

There are multiple levels and scales of interest, from low-level features such as texture to high-level features such as composition. On the top of that there’s a need for data-augmentation to compensate for sensitivity (e.g. training with blurred, cropped, scaled, noised versions of the images)

In order to do proper image analysis you will need huge (both deep and wide) architecture which requires massive amount of memory and processing power, made more accessible today via GPU empowered machines. It still takes a lot of time (up to 10 days) to train some large architectures.

There are a few implementations for convolutional neural networks.

  • cuda-convnet - python interface, c++/cuda implementation for convolutional neural networks, training using back-propagation method, fermi-generation nVidia GPU(GTX 4xx, GTX 5xx, or Tesla equivalent) is required, but no multi-GPU support
  • cuda-convnet2 - an upgrade to cuda-convnet, optimized for new kepler-generation nvidia gpus and added multi-gpu support
  • caffe - deep learning framework, developed by Berkeley Vision and Learning Center, big community of contributors, support for nVidia’s GPU accelerated convnet library - cudnn
  • torch7 - lua interface, with support for python, wide support for machine learning algorithms, one of the fastest implementation for convnet is a torch7 extension - fbcnn from facebook artificial intelligence research team; there are other extensions as well and a support for cudnn
  • theano - a python library, open-ended in terms of network architecture and transfer functions, slightly lower-level than the other implementations

We will be doing a series of articles in our blog on how image recognition is changing the paradigm and will allow image intensive business to finally better understand and monetize their image contents.

[slideshare id=47160420&doc=imaggaproductpresentationmleventslideshare-150419060621-conversion-gate02]


Apollo App Uses Imagga APIs to Help People Get Current Imagery from Places of Interest

Apollo App

Photo hack days are great fun and we are always surprised by the creativity and sometimes crazy ideas hackers have. Even though we were not physically present this time.

The runner up for Imagga API award was team Apollo, named after NASA’s Project Apollo, which gave us the first images of man walking on the moon.

"Our team began with the premise that mainstream photo sharing is characteristically narcissistic. As photo sharers, we often tie our own self-worth to attributions of our digital self. If we could make photo sharing altruistic, there's a chance that an entirely new set of content would be created", shares Joe, one of the guys behind the idea.

On AirBnB, public attraction and school websites we see polished images that sometimes do not correspondent with the current conditions of the place. Quite disappointing to head to a well deserved vacation just to discover your hotel hasn’t been renovated the last decade and the photos based on which you booked are heavily photoshopped.

The solution allows users to request a photo from other users that are currently present in a specific location. All users in that location would then receive a notification requesting a photo, and all of the photos for a given location are respectively tagged and stored in an image database.

Apollo Team presenting at Photo Hack Day NYC
Apollo Team presenting at Photo Hack Day NYC

Photos are curated by location (utilizing FourSquare's places API) using both user requested photos and those from existing content sources (e.g. ShutterStock, Behance, FourSquare). Imagga image recognition API was used for photo tagging and Twilio was implemented to make a photo request via SMS.

As a hypothetical use case, think of a prospective college student that wants to obtain a more realistic view into dorm life at the universities they’re considering. The student could request photos from dorm room locations at their target schools and current students share a photo of their room. It can also be used for inspirational purposes - say your are organizing a party at a hall and want to see how other organizers used the space and to get ideas for your own party.

"Imagga’s image recognition API quickly became a huge asset to us as we built out Apollo, enabling our team to intuitively apply relevant tags to the photos that comprised our image database.”

You can find more detail about Apollo App on Challenge Post.

We are really looking forward to see how the idea unfolds and becomes a real project.

There are thousands of ideas that Imagga API can help you make reality. Sign up for Imagga API and start hacking for free (or send us a request for special startup discount for bigger API volume)


Imagga to partner Photo Hack Day NYC

Photo Hack Day NYC

It’s already a tradition to take part as partner at Photo Hack Day, organized by Aviary. Photo Hack Day this time is in New York, and is organized for 4rd time (Imagga partnered the second and third editions of the event in Facebook HQ, Melno Park and in Tokyo, Japan). The hackathon brings together talented developers and designers to build amazing image applications using web and mobile photo APIs. This edition is sponsored by Shutterstock, with API partners like Walgreens, Astra by Photoshelter, Foursquare, Filepicker, Twilio, Behance, Shutterstock, Aviary and of course Imagga.

We are giving free usage of our APIs for all participants during the hackathon, 6 months free of our Pro plan (worth $2,094) for the best use of our API & 6 months free of our Indie plan (worth $474) for all teams that use our API.

Here are some good idea how our API can be of help:

  • Personal photo organization app based on auto-suggested tags or categories
  • Twilio-powered app where people can automatically text SMS/MMS by taking a picture: for example taking a photo at the beach and sending it with auto-text "Having fun under the sun :)" or a concert photo with auto-text "the band is killing it!", etc.
  • Dropbox/Box/Copy app that tags your photos and make them searchable
  • Pinterest image analytics tool that tag or categorize all images from an user/pinboard/whatever and shows the distribution of the different type of content, for example 15% dog photos, 19% cat photos, 65% horse photos, 1% pizza photos

It’s awesome to take part in photo hack meetups. Will keep you posted about the results and the great implementations as result of the hack day.

Photo Hack Day API Partners


Imagga @ Telerik Hackathon

Georgi Presenting @ Telerik Hackathon

After the exciting news of Telerik being acquired for over $260M last week it feels extremely exciting to report Imagga's API partnership at Telerik Hackathon. Being part of developer events like that reminds us of why we actually do what we do. Feels rewarding.

Imagga APIs help developers to quickly master any image data and extract meaning with the speed of light. Uses are numerous - you can extract color information and implement color search, classify images in predefined set of categories, extract keyword tags out of not annotated photos, even get the system learn from user feedback and improve  the recall.

Telerki Hackathon

Georgi presented Imagga API's during the opening of the hackathon. Over 200 eager developers got excited about the awesome ideas they were about to build during the weekend. Three of the formed teams requested access to the API and it was our pleasure to walk the teams through the APIs.

Being part of Telerik Hackathon perfectly aligns with what we at Imagga intend to do. Reaching out to the dev community is important part of our strategy to make image recognition mainstream and showcase how Imagga technology can be used to solve real world problems and help harness big amounts of visual data that bombards us on a daily basis.


Imagga @ Seedhack lifelogging hackathon

Georgi SeedHack Imagga
Copyright @seedcamp

It has become a good tradition to take part in Seedhack hackatons. This is our 3rd one as API partner and we can’t be more happy to support and learn from events like that.

This time the theme was lifelogging (previous themes were hacks for fashion and remixing content) and gathered around 100 eager hackers on Friday night. The lifelogging topic is still under-explored but there’s a lot of promise, especially with giants like Apple and Samsung investing heavily in not just software but hardware to advance this trend.

Friday night is exciting for the partners of the event - they get their 5 min on stage to present what they have in store for the event. Facebook did present their back-end platform Parse, PayPal - integration options including BrainTree, and Georgi from Imagga our new and hot API for automated image tagging along with the powerful image categorization API.

The rest of the event by tradition belongs to the hackers itself. 17 teams have been formed and got their hands dirty on solving world problems. It’s alway awesome to meet and talk with minds so bright. About 15 people came to talk about our APIs and different ideas they have either for the hackathon, or about something else they do in their jobs or projects.

It’s amazing how much people are able to achieve in 48 hours when they work focused on a problem and collaborate in a team. So much you need to consider - get the right people on the team, with good moral and excellent skills, quickly pass through the technology stack, feature list discussions and start hacking, then validate your ideas, pivot and validate again, talk to people, incorporate their feedback, ignore some, hack more, prepare a deck, rehearse it, get pumped for the public presentation, win.   The team that used our auto-tagging was Snipik - social review blogging platform. The winner of the hackathon is Calrity - "a smart calendar client that understands your context, gives you insight into how you spend your time and seamlessly organizes your future" - amazing progress implementing a complete prototype in just two days. They’ve also wrote an amazing blog post for Seedhack unveiling the whole experience during the event.

One of our favorite teams, Dev Stats won the special prize. Dev Stats is a keystroke counter plug-in for Sublime (Nike+ like tracker for programmers, geeks and nerds ;) They also won the "most likely to catch you slacking at work" award.

Some funny awards followed. The "dirty dancing" award went to Motion perfect - Kinect-based software to detect your fitness moves and help you control the quality of your workout. The "Tony Stark award" for highest achievements of a one-man army went to Popular - logging your offline social life. The "most likely to wake up a sleeping baby" presentation award went to Mr. Zen - advice from voice based mood logging.

Hackathons are one of the best type of startup events to attend and also partner. You meet with awesome people, movers and doers, get inspired and go home with hope and urged to move things faster in your own company. Definitely will be attending some more hackathons. See you there!


Imagga @ LDV Vision Summit

Imagga @ LDV Vision Summit in NCY
The LDV Vision Summit brought  together world’s top imaging technologists, visionaries and investors to share and shape the future of image and video recognition technologies.

Organized by Evan Nisselson, LDV Capital the summit was held in NYC to showcase how imaging and video technologies are empowering and disrupting the our day to day life and global businesses.

Visual web is upon us. Images and video are starting to dominate all web channels we are collecting data and impressions from. It’s not just the digital cameras and smart phones, but also drones equipped with cameras that will basically record and archive important moments.  This will  eventually enable us to take part in moments, conversations and event we couldn’t attend or be physically present.

The advance in GPU power, machine learning algorithms, well labeled data and some magical tricks implemented by various researchers and startups working on image recognition will eventually create new ways of exploring the visual web  - collection of personal and common image and video experiences.

It was exciting to see Georgi presenting Imagga on stage of LDV Vision Summit as one of thoroughly selected startups developing image recognition products and services. On stage were as well  Eugenio from  teradeep,  Alex from visiosafe  and Chris from TAG Optics. Quite good company. Some of the exciting lectures at the event were Camers - Video, Still & Wearable with panelists Martin Kallstrom from Narrative and Flo Meissner from EyeEm. The panel on Recent Progress in Computer Vision using Deep Learning with Rob Fergus from Facebook artificial intelligence lab was also quite interesting as it tackled the technology challenges we are facing with Imagga’s image recognition technologies. The Future of Image Recognition discussion session with Serge Belongie from Cornell NYC Tech,  Gary Bradski from Magic Leap and Moshe Bercovich, Shutterfly set the trends for the next couple of years and was followed Trends in Visual Communications Technology from the investor’s perspective with the awesome Om Malik from  True Ventures and Evan Nisselson.

Imagga’s automated tagging technology is what we consider groundbreaking and with huge potential to contribute to the change we perceive image  content in the near feature. According to KPCB Internet Trends 2014 report over 1.8 billion photos are uploaded and shared on a daily basis. Most of image date is not properly keyworded or not tagged at all. Memories are stored in archives, in the best case somewhere in the cloud. Exploring this visual records of our lives, interactions, personal encounters and important moments will be of key importance in the years to come. Using machine learning and super smart image analysis, technologies like our tagging API will be able to bring back the meaning to important moments, shorten the time we dig for them, automatically highlight the best quality content and feature significant events.

Beside the imaging summit, being in NYC is always exciting, especially for image intensive business like Imagga’s. So many of our current customers are headquartered in NYC, so couple of extra days were well spent on biz dev meetings. Thanks to the summit we’ve got quite valuable new prospects and couple of partners.

 


Photo Hack Day Japan - Imagga Prize

Photo Hack Day Japan

It’s exciting to take part in photo hackatons. We’ve been to several during 2013 - Photo Hack Day Berlin, Photo Hack Day Melno Park, SeedHack in London to name a few. This time we were not present physically but that did not prevent people from using Imagga APIs.

Over 120 hackers and supporters gathered on Feb 22-23rd at Mixi’s office for an awesome hacking experience. Sponsors of the event were Mixi, Shutterstock, gettyimages, 500px, Leap Motion, Amazon, Mashup9, Recruit, EyeEm, Pux, XLoudia, Sendgrid, Zenrin, FlashFoto and Adobe and we at Imagga are quite happy to be listed among this great innovative companies.

As any serious hackathon it all starts with opening remarks of the organizers and API partners presentations just to be followed by the creative chaos of team formation, idea generation and hacking over lots of fried rice, shushi (it’s Japan after all) but also coffee, tea and beer. It’s great fun and we were a bit jealous we’re not physically present.

Photo Hack Day Japan work

23 teams have been formed and were judged based on originality of the idea, overall value of the application and technical achievement. The tree overall winners of the hackathon are:

  • Back to the Future - visual learning app that allows you to "walk through time” via an interactive interface.

  • Before the filter - Android app that teaches would-be photographers the fundamentals of Photography

  • VOCA Getty - innovative photo dictionary app that uses images sourced from Getty Images to help users learn a language visually. The team used Imagga APIs.

We were excited that 3 teams actually activated, reviewed and used Imagga APIs during the event.

photo tagging

Our special $2299 in credits prize goes to  Imagic Workbench, a smart keywording tool which employs object recognition technologies to automatically keyword your photos. Imagic Workbench is a project of TT and Nego and we really hope they will follow up and make an awesome product for image tagging.

Stay tuned for our next hackathon participation, it's fun!

Interested to try our auto-categorizatiom? - signup for a free account at our Playground


Imagga @ Photo Hack Day Japan

Photo Hack Day Japan

Imagga continues to take part in hackathons. It’s in our DNA and it’s one of the best ways to show support for the photo hacking communities all over the world. This time Photo Hack Day 5 hackathon will take place far far away of where we are. Japan is exciting place for tech events not only for the level of tech penetration but also for the thriving creative community engaged in visual arts.

Thanks to our friends from Aviary we will be partnering Photo Hack Day, that will take place on February 21-23 at Mixi Headquarters in Tokyo. Hackers will be able to play with Imagga API's and specially with the recently announced Imagga Auto-categorization API.

It's too bad we will not be able to make it in person for the great event, that's why we produced short video that describes what we do and how exciting Imagga APIs are.

Here are some ideas on how Imagga APIs can be used:

  • Our brand new auto-categorization API can be used for personal photo sorting. But you can also train your own categorizer to recognize particular types of objects or scenes or even moods depending on the use-case. Let say you want to build a Pinterest like site but for cool products, manually crafted locally. In most of the cases you will be able to scrap the photos from personal blogs, artist online shops but they might not be well structured and organized. You can train the with selected product categories our APIs and then all photos will be automatically recognized and placed in the relevant category.
  • you can create some nice looking retrospective photo album with the photos smartly cropped, something like a better version of the Facebook Look Back video, but without bodies and faces cut in half
  • what is your favorite color app that can detect your favorite colors from the clothes that you wear or things you posses
  • detecting mood by color, or suggesting music by color apps

We will award $2 300 in Imagga API usage for the best use of our APIs and $400 in API usage for all teams who make use of them during the hackathon.

You might also have great ideas about possible implementation of Imagga APIs. Let us know about your idea and we will be helping out with any technical challenges you might have.


Imagga @ leWeb 2013

Chris & Guy Kawasaki

leWeb is one of the biggest and oldest European event for entrepreneurship. It’s always been well attended by entrepreneurs from all over Europe and beyond. More than that, people the startupers admire - we read their books and blogs - are actually speakers at leWeb. And they keep coming every year. After Techcrunch Disrupt Belin and Web Summit in Dublin, leWeb wasn’t in our program for this year, but Chris won a free ticket from Invest in France and that was a good sign we should attend.

It was exciting to be there and see the French way of doing events. It’s relaxed but at the same time quite exquisite. Techcrunch Disrupt has the hipster flavor of Berlin, Pioneers Festival enjoys the most luxurious venue being held in Hofburg Palace in Vienna, WebSummit pretends to be the biggest event in Europe, while leWeb is just French, and it’s French in a very good way - extremely good food all day long, comfy lounge zone to meet people and charge your devices, the best lineup of speakers ever.

Best events are not just what happens on stage, but also the networking. With parties every night leWeb offers lots of opportunities to meet people. Well, meeting investors was quite a challenge as they were kept in a VIP area just above our heads (a bit irritating to segregate and make it so visible even during parties ;-) ). Do not ask how we ended up there... friends ;-)

Need not to say we spent most of the time networking, not just during parties but during the day as well. Well, couple of the speakers were must listen as these guys are setting trends and always have so much insight on the whole ecosystem so we decided to better sit and listen.

One of these people we really enjoyed listening was Fred Wilson. He was invited to speak about what’s going to happen in the next ten years but instead he shared 3 big trends he’s interested investing in:

  • Transition from bureaucratic hierarchies to technology-driven networks. Think Twitter vs. classic newspapers, YouTube vs. cable tv, SoundCloud vs. traditional music labels. We are in the early stage of that trend, but it’s happening not just in media, but hotel reservations (AirBnB), creative (Kickstarter), learning (Codecademy, Duolingo)
  • Unbundling - deliver much more focused services vs. the old model of newspaper with business pages, culture, sports section. This is already happening in banking, education and entertainment.
  • Smartphones - people are more willing to sacrifice their laptops rather than smartphones if they need to choose one. With the phone you can do things you can’t with your laptop and there will be way more great applications to utilize mobile. Great examples are apps that are disrupting taxi service (Uber), payments (Square), dating (Tinder)

He stated he doesn’t look at technologies but in trends, that will have behavioural and social impact. Of course mobile, big data and machine learning are hot technologies but it’s not technology itself, but it’s practical implementation. Can’t agree more working on elaborate machine learning technology such as Imagga’s image auto-categorization.

You can watch Fred Wilson’s video here:

http://www.youtube.com/watch?v=R43OKYmGbhU

Another much expected talk was Guy Kawasaki’s - a legendary role model for the Bulgarian entrepreneurs. What a precious moment to have the chance to chat for a minute with Guy and get his best selling book The Art Of The Start signed by him.

Chris_Kathleen_Stephane_leWeb

leWeb was eye-opening experience. Lots of contacts to follow up, great time networking, inspiring speeches and memorable meetings with awesome people. Special thanks to Stefan from LAUNCHub for hosting us and being the perfect guide for cool places in Paris (experiencing Uber on his account was awesome ;-) )


Microstock Expo in Berlin 2013

Microstock Expo Berlin 2013

It’s our second visit to Microstock Expo - the best European event for stock photography. We visited the first year the conference was organized and all the lectures were extremely insightful.

Stock photography is slowly changing. Not only professional cameras are quite accessible for the masses but the rise of the smart phone photography gets even more people excited about the possibility to make money even with photos shot with not so good cameras as current mobile phone cameras are.

There were couple of lectures that were quite relevant to what we are up to currently in Imagga.

One of the most interesting topics during the conference was the keywording. What was striking for us that most of the people from the industry do not actually believe the keywording can be really automated. The human touch still plays significant role in determining the key words that best describe an image. It’s probably needless to day we totally believe there’s room for automation in the keywording. Obviously we have the difficult task to convince partners that we can really be of help with technology like automated tagging.

MIcrostock Expo Berlin

One of the most practical keynote speeches was done by Luis Alvares from Stock Performer, a platform for micro stock analysis. They’ve done quite extensive analysis of the words that sell well and once that are quite widespread but actually do not sell well at all.

Overall - great event with what of potential partners, so we are excited we were part of it.