Google continues to push the AI envelope with release of new toolset that lets robots 'see'
By Michelle Goodman
With each announcement of an artificial intelligence breakthrough, smart robots become less of a futuristic movie plot and more of a reality integrated into daily living.
Google's beta release of Cloud Vision API, which has been in limited preview since December, is the latest in a recent spate of AI announcements from software giants. Microsoft, IBM and Yahoo have also made impressive AI strides in recent weeks.
With Google's new AI toolkit, developers can use machine learning to help robots recognize image content. Hold up a banana, and the bot can identify it by name. Program the bot to move toward smiling faces and away from frowning ones, and it will recognize facial expressions accordingly and move in the desired direction.
By using the technology behind Google Photos, Cloud Vision API can detect a wide swath of objects (think flowers, food, animals and notable landmarks). Other possible real-world applications include detecting inappropriate content in crowdsourced images via Google SafeSearch, detecting popular product logos, analyzing the emotions of people in images and identifying text and languages in images via optical character recognition, according to the Google Cloud Platform Blog.
Anyone can submit their images to Cloud Vision API during this unspecified beta timeframe. Each user can submit up to 20 million images a month, with pricing varying depending on content detection needs and image volume.
Thousands of companies have used the API since December to generate millions of requests for image annotations, Google said. For example, PhotoFy, a social photo editing and branding app that moderates more than 150,000 crowdsourced photos daily, has been using the API to weed out potentially violent and pornographic images.
Pichai is making good on his promise to step up AI efforts at Google. And the fact that the company is jockeying with Microsoft and IBM to create the best machine learning tools is a boon for AI developers, not to mention robotic innovation overall.
But it's racing against others, including Microsoft, which released a new set of machine learning tools to GitHub, proclaiming their superiority over those of the competition. It has also released tools that can recognize emotions based on facial expressions in photos and identify dog breeds.
Google shares its machine learning software, TensorFlow, with the world
Machine learning key to 'rethinking everything' at Google, says CEO Pichai
Microsoft releases machine learning toolkit on GitHub in another win for AI research
Apache Incubator accepts IBM's SystemML for open source development
Yahoo brings machine learning data to the research community with huge dataset release