Google Cloud AutoML Lets Users Create Smart Image-Based Apps
Building on its vision to make AI "work for everyone," Google today launched a new cloud-based service for building artificial intelligence systems without advanced programming expertise. Cloud AutoML is aimed at users that could benefit from AI-powered applications for business, research, conservation, and other needs.
The first tool Google is rolling out under the new service is Cloud AutoML Vision for intelligent image recognition. Rather than requiring code, AutoML Vision uses a drag-and-drop interface to help users train, manage, and deploy machine learning-powered applications for image-based intelligence.
For example, a clothing store could use the service to help customers search for and find tops with scoop necklines without having to first manually tag every individual product with that style. Instead, Cloud AutoML Vision could be used to "train" the retailer's systems to automatically recognize scoop-necked tops based on product images alone.
On a Mission To 'Democratize' AI
During Google's I/O developers conference in May, CEO Sundar Pichai highlighted the company's goal of helping customers apply AI and machine learning to a broader array of daily tasks ranging from email and job hunting to healthcare and virtual reality. One of the initiatives Pichai unveiled as part of the mission to "democratize" AI was an effort to use machine learning to design even better machine learning.
That capability is an essential part of what Google's new Cloud AutoML does: it uses advanced techniques, such as transfer learning, which takes knowledge gained from working on one type of problem and applies it to different types of related problems, or automated learning, that is, teaching machines how to learn to learn.
"Currently, only a handful of businesses in the world have access to the talent and budgets needed to fully appreciate the advancements of ML and AI," Google cloud AI chief scientist Fei-Fei Li and cloud AI head of research and development Jia Li wrote in a blog post today. "There's a very limited number of people that can create advanced machine learning models. And if you're one of the companies that has access to ML/AI engineers, you still have to manage the time-intensive and complicated process of building your own custom ML model."
Cloud AutoML will enable even organizations without advanced machine learning skills to use Google's intelligent tools to "build powerful AI systems they previously only dreamed of," they said.
Other Services Could Include Video, Speech
The first service available through Cloud AutoML can help users build and launch smarter, more accurate vision-based applications that are easy to use and quick to deploy, the Google researcher said. Early adopters of Cloud AutoML Vision include Disney, the Zoological Society of London, and the clothing retailer Urban Outfitters.
In their blog post, Li and Li quoted Alan Rosenwinkel, a data scientist with Urban Outfitters' parent company, who said, "[M]anually creating product attributes is arduous and time-consuming. To address this, our team has been evaluating Cloud AutoML to automate the product attribution process by recognizing nuanced product characteristics like patterns and neckline styles. Cloud AutoML has great promise to help our customers with better discovery, recommendation and search experiences."
The Zoological Society of London, meanwhile, has been using AutoML Vision to automate the tagging of millions of images of animals in the wild taken by heat- or motion-activated cameras.
Google is currently developing several other AI-powered services within Cloud AutoML, according to Li and Li. While they did not provide details on what those services would cover, such applications could cover such areas as video, speech, text analysis, or translation.
More than 10,000 businesses are already using other Google Cloud AI services, including the cloud storage provider Box and Rolls Royce Marine, Li and Li noted.
Image credit: iStock/Artist's concept.