google

Google has acquired France based machine learning startup Moodstocks for an undisclosed amount, the company said in a blog post. Post acquisition, the entire team of Moodstocks will join Google’s R&D facility at Paris, and they will help strengthen visual recognition capability in Google’s image or video based products.

Although Google doesn’t exactly hint on products that Moodstocks will be working on, it said that the team will join researchers and engineers who had earlier contributed to development of YouTube, Chrome, and Cultural Institute at Google’s R&D centre in Paris.

Moodstocks mentioned on its website that it launched an “on-device image recognition” software in 2012, which “turns cameras into smart sensors able to make sense of their surroundings”. Currently, the company works with visual recognition algorithms, machine learning, and an image recognition technology on smartphones to help users identify objects.

Google already does what Moodstocks used to do

A YouTube video shows an iOS app developed by Moodstocks identifying and naming different objects from small mugs to furnitures, with an option to buy them online: (P.S: This was first spotted by The Verge)

Interestingly, Google already has a similar mobile app named “Google Goggles” that does a range of image recognition function like identifying famous landmarks through an image, or fetching product information through its bar code.

In addition, Google mentioned that machine learning is already present in products like Google Translate, Smart Reply Inbox, and Google app. It also highlighted about image recognition capability on Google Photos: a user can search for keywords like “party” or “beach” while the app fetches images taken in a party or near a beach. Much like Apple’s facial/image recognition feature that was unveiled on iOS 10 during Apple WWDC developer conference held last month. More on that here.

Google isn’t the only one investing in machine learning

-Twitter: Last month, Twitter had acquired London-based Magic Pony to enhance its machine learning capabilities in image processing; this was Twitter’s third acquisition in a machine learning company. Magic Pony’s technology, which creates algorithms in visual or image processing, will be used to enhance its live video product Periscope, short clips product Vine and other video and image based features.

-Apple: In October last year, Apple had acquired machine-learning firm Perceptio, a startup that worked on advance image-recognition capabilities using artificial intelligence, especially on smartphones. This later yielded in iOS 10 image-recognition feature, which we highlighted earlier. Last month, Apple introduced a new facial recognition engine within iOS. It lets users identify friends and family members in photos. iOS can now detect objects, scenery, mountains, water, locations, etc, and users for example can search for “Pics taken near mountains”.

Facebook: In January 2015, Facebook had acquired Wit.AI, a San Francisco-based startup that developed a speech-recognition platform for different devices and offering it through APIs. This later powered Facebook’s machine learning capabilities for text-translation on its social media website; earlier Facebook depended on Bing for text-translation, but ditched it after acquiring Wit.AI. Currently, Facebook claims that it can translate across 40 different languages in 1,800 directions (for e.g. French to English), and also claims that over 800 million users, have already tried it out.