Developing an artificial intelligence (AI) system probably sounds overly complicated for most individuals. After all, AI systems themselves are highly advanced, so conventional wisdom may lead you to believe that creating one is equally if not more difficult. Well, Google has launched a new platform to hep users create custom AI systems.
Known as Cloud AutoML, the platform uses machine learning to develop advanced AI-powered image recognition systems. According to Google, such systems typically require tens of thousands of images and a special recognition program. However, Cloud AutoML allows users to create their own custom image recognition systems using just 100 images. By lowering the barrier of entry, Google hopes to provide AI-powered image recognition software to countless businesses and entrepreneurs throughout the world.
So, how is Cloud AutoML able to create image recognition systems using just a small sample of images? It achieves this feat by leveraging cloud computing technology. With images stored on the cloud, AI systems can automatically cross-reference sample images while looking for a match. This eliminates the need for storing images locally, which would otherwise consume more computing power as well as storage space.
As explained by MIT Technology Review, Cloud AutoML is currently limited to a select group of customers. However, Google says it plans to expand the platform in the near future. The price for Cloud AutoML is unknown at this time.
Of course, Google isn’t the only company that offers AI training tools for image recognition. Amazon and Microsoft both other similar platforms but with limited functionality. Google’s Cloud AutoML is a more user-friendly tool for creating AI systems.
Google has been working aggressively to expand its footprint in the AI market. In 2017, Google CEO Sundar Pichai announced a new division dedicated to AI research. Known as Google.ai, it’s helped to create cloud-based tensor processing units (TPUs) and other AI-powered technologies.