5 Computer Vision and Image Understanding Companies
You’ve probably heard by now that Google’s artificial intelligence program called AlphaGo beat the world Go champion to win $1 million in prize money heralding a new era for AI advancements. Go, which has been played for 2,500 years, is difficult for a computer to master because it has more moves possible than atoms in the universe. While the sad news of the Korean champion being dominated by a computer caused many Koreans to drink heavily, observers of this feat thought it remarkable that AlphaGo was actually using strategies that had never been used before by humans.
This deep learning program was first fed about 30 million moves to mimic, then it was allowed to play against instances of itself in order to “learn” using a concept called reinforcement learning. It’s kind of similar to how you train animals to do tricks for a circus. While some of the concepts used like Monte Carlo are much easier to grasp, the program itself uses neural networks that mimic how the human brain works. One of the more popular uses for “conventional neural networks” is in the applications of computer vision and image understanding. In fact, computers are now officially better than humans at image classification:
We talked before about how you can train a computer to recognize images by looking at them instead of just reading the text that accompanies the picture. Here are 5 startups that are playing in this space.
Founded in 2014, French startup Deepomatic has taken in nearly $1 million to build a technology that will enable content publishers to monetize their images by linking them to purchasable products. You know that whole “a picture is worth more than a thousand words” thing people say? Here’s the best way to describe Deepomatic’s business model:
So there you have it. A perfect technology for selling stuff to people who idolize talentless celebrities and want to dress just like them. They’re presently looking for a “Chief Fashion Curator”. If you think you have what it takes for this technically demanding role, just apply here.
Founded in 2009, this Massachusetts startup has raised nearly $20 million from investors that include Kleiner Perkins Caufield & Byers and Horizon Ventures. Affectiva has analyzed nearly 4 million faces across 75 different countries to build a program that in real-time can analyze any of the 10,000 possible facial expressions you could be displaying at any given time. Want to see how it works? They have an online demo you can try which will analyze your facial expressions through your webcam while you watch humorous Doritos adds. Considering that 90% of people walking around have their faces glued to their smartphones, it’s not hard to conjure up the large number of possible commercial applications this technology can be used for.
Founded in 2014, Silicon Valley-based startup Hyperverge has taken in around $1 million in funding to develop image recognition technology for the identification of people, places, scenes, events, documents and objects within photos. With all the team having met in school at one of India’s top colleges, their “About Us page” looks more like the corporate directory of your average investment bank’s back office except none of them look old enough to drink:
Their first product is a photo sharing app called “Silver’. With over 285 million images recognized and growing, Hyperverge has also built technology that analyzes the quality of photos to identify the best photos, bad photos, duplicates, spam photos, etc.
Founded in 2014, London startup Tractable has taken in $1.9 million in funding to develop “proprietary machine learning algorithms, with a focus on deep learning for computer vision“. Their one-page website doesn’t contain too much detail but does let us know that they’re targeting auto insurance claims as one of their first applications. One of Tractable’s advisors is the former CEO of Crawford & Co which is a branch of the world’s largest independent claims management company. Several other applications they are targeting include remote monitoring and industrial inspection, two application areas where it’s pretty easy to visualize how computer vision might be used to
take away jobs free up people so they can be used to work on more value-adding activities.
Founded in 2013, this New York-based startup has taken in $10 million in funding so far to develop an AI program that will take any image you give it and tell you attributes about that image. You could feed it all your travel pictures for example and then start asking it to show you things based on natural language queries like “show me all pictures of my family at the beach”. At some point, we wondered just how much money you can make selling people a service that will classify their images for them. Clarifai offers this service as an API with the following subscription costs:
So this business model works well but will eventually dwindle down to a single company that offers this service better than anyone else so everyone adopts it. It lends itself to the same kind of outcome we saw when search engines first came out. The end result is one company that dominates the space while the leftover companies die slow deaths while switching CEOs faster than most people switch jobs. Speaking of which, here’s what Clarifai comes back with when you feed it a picture of Marissa Mayer:
Are you paying too much in transaction fees to your broker? Check out a brokerage firm called Zacks Trade that's offering $1 trades on U.S. stocks and options until 2020. After that, you'll pay just $3 a trade or a penny a share, whichever is greater. You can also trade on 91 stock exchanges in 19 foreign countries. Click here for $1 trades on U.S. stocks and options until 2020.