ABI Research: Machine Vision Boosted by the Increasing Use of AI Chips in the Mobile Devices Market

Shipments of mobile devices, including smartphones, tablets, and wearable cameras, with embedded machine vision technology, will increase from 45 million in 2018 to over 590 million in 2023, according to a new forecast by ABI Research. That amounts to a combined annual growth rate (CAGR) of 67 percent.

Mobile devices, particularly smartphones, are increasingly featuring embedded machine vision technology, which offers a wide range of visual applications that scan, analyze, and interpret images or video.

“Smartphones offer a convenient form factor for machine vision, and with an increasing number of Artificial Intelligence (AI) chips in use, such as Apple’s A12 Bionic and Huawei’s HiSilicon Kirin 980, its prevalence is likely to increase,” said Stephanie Tomsett, Research Analyst at ABI Research. “For devices without an AI chip, as long as the smartphone has a camera and cloud connectivity, the technology can still be leveraged.”

Machine vision is mainly being used by mobile devices for image recognition, providing users with detailed information about an object, which can include a building’s history, an item’s location, and a flower’s name.

Other use cases include automatically detecting issues with crops in agriculture, automatically determining the course of treatment for a wound in healthcare, and automatically detecting issues with a piece of machinery in manufacturing.

Consumers are increasingly taking advantage of the technology, allowing them to search for information on what is viewed through a smartphone, without having to type in any search criteria.

Machine vision technology uses in-built image recognition algorithms to identify the object within the image, which is being increasingly powered by an AI chip. Companies such as Apple are providing these chips on their latest iPhones, with others such as Huawei offering them on their flagship products.

For devices without an AI chip, the machine vision process is done via the cloud, allowing any device with internet access and a camera to utilize the technology, using algorithms from companies such as TensorFlow.

“For mobile device vendors looking to offer on-device machine vision, a number of important factors must be taken into consideration, including the hardware required to power the technology, the privacy laws in place, and the available datasets. In the next two to three years, it is likely that many major smartphone manufacturers and limited tablet and wearable camera vendors will offer embedded machine vision to the higher-end model lines, enabled by the integration of AI chips,” Tomsett concluded.

These findings are from ABI Research’s Machine Vision in Mobile Devices application analysis report.

(For more information visit www.abiresearch.com).