Free and open-source software (FOSS) refers to software which anyone is licensed to use, copy, study or change in any way. This model of licensing makes sense in a number of ways. On the pratical hand, it allows a broader user base to contribute, enhance and exapand a project, and also to monitor and improve its security. On the philosophical hand, it allows for more freedom to play and to tinker, the liberty to use a piece of software as one sees fit, as well for a much higher degree of control.
At Zeta Motion, we pride ourselves to be operating based on the principle of privacy by design and to adhere to a strong committment to ethical innovation, both technical and and business-wise. In our mission to democritise motion capture (MoCap), it was clear from the very beginning that we would try to provide as much freedom and power to our users and customers as possible. To provide open source tools to make use of our core technologies was therefore a natural step.
Today, you can find part of our effort in that direction on github, with Minimind2. Originally conceived by our CTO, Dr. Huy Phan in 2017 as Minimind1, it has just been updated to version two and will find its application in various of our products and services. In technical terms, it does the following:
Minimind2 is a minimalist machine learning library written in Swift 5.0. It allows developers to quickly prototype algorithms by providing a friendly numpy-like interface and native performance on platforms such as iOS and macOS. Designed around Swift’s extension paradigm, Minimind2’s Array and Matrix structures support a wide range of data types, including Float[16, 32], Double, [U]Int[8,16,32,64], Bool and even String. Minimind2 automatically makes use of hardware acceleration whenever it is available via Accelerate, vDSP and vImage libraries.
In layman’s terms translates to: Minimind2 allows developers to quickly prototype algorithms by providing a concise and friendly interface while making use of hardware acceleration whenever it is available.
Check it out at github.com/ZetaMotion/Minimind2