03 Nov 2015 Clarifai
Every day it becomes clearer that artificial intelligence (AI) is the future of technology. The giants of the tech and Internet industries—Google, Facebook, Apple, etc.—are all hiring AI experts and purchasing related startups. Image analysis is a growing area in this field.
Clarifai is one interesting startup focusing in AI for image analysis. They have a solid base of experts at the helm, and they’re providing an API for their service at very affordable rates. With this API, a lot of microstock agencies are about to add auto-tagging functionality to speed up and improve image keywording and visually-based similar search.
AI for Visual Analysis
Clarifai’s technology is based in artificial neuronal networks; the “ai” in their name stands for artificial intelligence. They develop and train deep learning algorithms in image recognition.
Their products for the stock industry include tagging (image keywording), similar search (either similar image suggestions on photo pages or reverse image search from photos uploaded by customers), and data collection and analysis on images’ visual appeal or suitability.
Experts at The Lead
This young company was founded in New York City in 2013 by CEO Matthew Zeiler and founding CTO Adam Berenzweig.
Zeiler received his doctorate in Computer Vision and Machine Learning from NYU under the supervision of Rob Fergus, an AI expert who was soon snatched by Facebook. He also spent five years working with the “godfather of AI development” Geoffrey Hinton, at Google, and with Andrew Ng, another eminent figure in AI also working for Facebook.
In 2013, Matthew won the AI competition ImageNet with the first version of his deep learning algorithm for image recognition. Around the same time, he rejected an offer to join Google’s team, choosing instead to focus in the independent project he had been working on, which would born as Clarifai later that same year.
On his part, Berenzweig holds a PhD in Machine Learning and Signal Processing from Columbia University, and he worked at Google for over 10 years, where he built Google Play’s music recommender.
The startup closed a series A funding round for a total of $10M in April this year. Union Square Ventures led the round, and Google Ventures was one of the funds taking part.
Today, the company’s staff has grown to around 20 people, including academics in software engineering and computer science and experienced software executives.
Clarifai’s deep learning algorithms analyze semantic (keywords and other metadata) and visual content to identify objects, themes and other elements in images. Through this analysis, the software can deliver identical and similar matches and suggest keywords and content.
They started offering image recognition for stills, but in January of this year, they launched the service for video content. This technology can identify themes and objects in a video and sort them by relevance as they appear, among other features. Clarifai also offers custom solutions upon request and claims to support up to billions of files.
These automated solutions for image search, categorization and analysis have very useful applications for stock imagery agencies. But the company’s services are also used in other industries.
An important point is that they deliver the services via their API, which was initially invite-only but rolled out to public this January, at very affordable prices. Sized by file volume, subscription tiers range from $19 to $479 per month, and they all include services for image and video.
Clarifai in Stock
Clarifai’s technology is probably what’s behind many of the smart functions microstock agencies have added recently, including contributor keyword suggestion tools and the upload-your-photo-to-find-similar-images function.
Their services and rates are perfect for stock photo agencies. If there are any microstock companies not yet working with Clarifai, won’t be long before they do.