At the MIDC 2019 conference organized by Xiaomi and held today, which in terms and name is very reminiscent of Apple's WWDC conferences, our Chinese giant has revealed several interesting information. The first of which is about Redmi K30 that almost certainly will arrive in December and that you have the opportunity to read , promising, while in this post we will discover the new version of the framework just announced by Xiaomi, the MACE 0.12.0.
Xiaomi presents MACE in version 0.12.0 at MIDC 2019
The MACE project, which stands for Mobile AI Compute Engine, aims to create a framework that uses artificial intelligence and deep learning to improve certain features especially in the photographic sector. Xiaomi has in fact given as an example the features that we find inside the camera app such as that of portrait mode (portrait mode with subject in focus and blurred background), that for the recognition of places, that to add stylized effects and images to very high resolution.
Furthermore, MACE would also be used for the translation of both text and images and for voice translation, but also facial recognition, voice recognition and much more.
Xiaomi has stated that MACE 0.12.0 will be open source, therefore usable by all, and will have support for various microcontrollers, such as those used in smartphones or IoT (Internet of Things) devices with low power consumption and low interference.
The strengths of MACE 0.12.0 will be mainly four:
- More compatibility in heterogeneous computing, a system that uses more than one type of core architecture to maximize performance and energy savings;
- Decrease performance losses between different products;
- Added Kaldi speech recognition system, an open source voice recognition toolkit available for free under the Apache license;
- Support for CMake, an open source tool for managing software creation using an independent compiler method.
Now we just have to wait for MACE 0.12.0 to be adopted by as many developers to see the technology grow ever more.
