Windows Vision Skills (Preview)
The Microsoft.AI.Skills.SkillInterfacePreview namespace has been deprecated as it is legacy and is no longer maintained. It has been replaced by non-monikored package offering the same functionalities and more with improve performance and reliability: Microsoft.AI.Skills.SkillInterface. This documentation is scheduled to be updated to reflect the new functionality.
Implementing and integrating efficient machine learning and computer vision solutions is a hard task for developers. The industry is moving at a fast pace and the amount of custom-tailored solutions coming out makes it strenuous for application developers to keep up. Existing APIs and lower-level frameworks add a steep learning curve before developers can leverage them effectively.
The Windows Vision Skills framework is meant to make it easier to utilize computer vision. It standardizes the way computer vision modules are put to use within a Windows application, running on the local device. It abstracts complexities into a single programming paradigm with standardized primitives, helping developers focus on building awesome computer vision applications.
The implementation that contains the complex details is encapsulated by an extensible WinRT API that inherits the base classes and interfaces in the Microsoft.AI.Skills.SkillInterfacePreview namespace. This API can be ingested by all types of Windows apps (UWP, .NET Core, and Win32). This framework is open for all developers to build on top of.
What is a skill?
In the context of Windows Vision Skills, a skill is a streamlined, modular piece of code that processes input and produces output. A skill can encapsulate simple functionalities like edge detection with a single-purpose micro-skill, or rich sets of functionalities forming a scenario skill to address a complex problem like skeletal detection.
Simple integration: Skills make it easy to add features to your application with out-of-the-box APIs that don’t require any machine learning or computer vision expertise, or prior Windows knowledge of low-level APIs.
Abstracting hardware acceleration: The Windows Vision Skills framework queries the hardware assets and provides OS provisions that allow developers to make efficient compute decisions at runtime.
Interoperability: The framework works with OS interfaces and assets such as image primitives from cameras, photos, and video, and it can be used in conjunction with the Windows Machine Learning APIs.
NuGet packages: Windows Vision Skills is strongly versioned to ease iteration without breaking existing applications. They are easy to ingest, easy to update, and they preserve intellectual property through licensing.
Extensibility: The framework can be easily extended to work with existing machine learning frameworks and libraries such as OpenCV.
Modularity: Skills can be pieced together in succession within an application just like a recipe to address a complex scenario. Skills can also be bundled together in a single package by the developer.
While this preview focuses on vision-oriented scenarios and primitives, the API is meant to accommodate a wide range of input and output variables that enable audio processing, text processing, and more.