Google launches AutoML Vision in bid to ‘democratise AI’

First of suite of products leveraging Google's pre-trained machine learning models and Neural Architecture Search technology

Google has launched a suite of easy-to-use tools that allow developers with limited machine learning expertise to train custom models.

AutoML leverages Google’s pre-trained machine learning models and Neural Architecture Search technology.

The products are part of Google’s “mission to democratise AI” and to “lower the barrier of entry and make AI available to the largest possible community of developers, researchers and businesses” the company’s chief scientist for cloud AI Fei-Fei Li and head of cloud AI R&D Jia Li wrote in a blog post yesterday.

The first product in the suite to be made available – to those that apply and are approved for access – is AutoML Vision, for building custom vision models based on Google’s proprietary image recognition technology.

Developers drag and drop images into a user interface in which they are able to label them, based on their requirements. The model is trained on the inputs, and users can then evaluate and further refine it.

Google is also offering “a team of in-house human labelers” that will review custom instructions and classify images accordingly.

Early adopters included fashion retailer Urban Outfitters who experimented to automate the recognition of nuanced product characteristics like patterns and neckline styles. Disney is using the technology to build vision models to annotate products with Disney characters, product categories and colors.

The Zoological Society of London has been using AutoML Vision to analyse and annotate images of animals caught by camera traps in the wild.

“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,” Li and Li wrote.

“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 and AI engineers, you still have to manage the time-intensive and complicated process of building your own custom ML model.”

Although Google introduced its Cloud Machine Learning Engine last year, delivered via APIs, the new suite will help “close the gap” and make AI “accessible to every business”, the pair added.

A vote for ML

Google’s cloud provider competitors are also on a push to “democratise” machine learning.

Microsoft’s Azure Machine Learning Studio is a “simple browser-based, visual drag-and-drop authoring environment where no coding is necessary” . In September the company expanded its AI tool range – although the products are still in preview mode.

Amazon Web Services, at its Re:Invent conference in November, launched a fully managed end-to-end machine learning service called Sagemaker and a video camera that runs deep learning models dubbed DeepLens.

“Builders don’t want machine learning to be so difficult. They don’t want it to be so cryptic. They don’t want it to be black box. They want it to be much easier to engage with,” said AWS CEO Andy Jassy at the time.

“There just aren’t that many machine learning expert practitioners in the world. Most end up living at the big technology companies. And if you want to enable most enterprises and companies to be able to use machine learning in an expansive way we have to solve the problem for making it accessible for every day developers and scientists,” he added.

Read more: Google says "no changes" to mapping platform in China after report

Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags trainingGoogledevelopersPreviewimage recognitionmachine learningMLAutoMLmodel

More about AmazonAmazon Web ServicesAWSGoogleMicrosoftUrban Outfitters

Show Comments
[]