According to a recent press release, “MediaTek today announced that in its continued collaboration with Google, the Helio P90 will support Google’s ML Kit, bringing machine learning techniques to mobile developers of all levels of expertise. MediaTek’s Helio P90, an AI processing powerhouse with advanced camera capabilities, will deliver software support to assist in making AI an essential and intuitive part of a developer’s toolkit. This news follows MediaTek’s recent announcement that it completed testing and enablement of ARCore and Google Lens on the Helio P90 chipset. ‘MediaTek continues to build a strong working relationship with Google to develop technology that is more optimized for mobile devices,’ said TL Lee, General Manager of MediaTek’s Wireless Communication business unit. ‘With Google’s ML Kit, we’re providing the right tools for smartphone users of all levels to apply machine learning to everyday apps to get the most out of your smartphone and continue bringing new premium features to the mass market’.”
The release goes on, “Google’s ML Kit is a mobile SDK that brings Google’s machine learning expertise to Android and iOS apps. With capabilities suitable for all experience levels, the ML Kit’s functionality can be implemented in just a few lines of code, or through the standard APIs for more experienced developers. To ensure that the ML Kit is properly supported, MediaTek adapted the ML Kit models including face detection to ensure that they can be accelerated on its APU 2.0. With software support from the Helio P90 chipset, Google’s ML Kit has seen encouraging results including 9x acceleration when compared to CPU implementation. MediaTek’s hardware works seamlessly with the Android Neural Networks API (NNAPI) with 80% of operations, including all compute intensive operations, supported by the AI accelerator. The NNAPI provides a consistent developer interface to allow more innovation across the mobile AI ecosystem. Plus, the ML Kit delivers ready-made solutions to core uses cases without requiring developers to have any knowledge of machine learning.”
Read more at PR Newswire.
Image used under license from Shutterstock.com