VisionWorks Toolkit Reference

December 18, 2015 | 1.2 Release

 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
Architecture

Provides an overview of the VisionWorks toolkit architecture and its components.

Introduction

VisionWorks starts from a fully compliant implementation of OpenVX 1.0.1 and provides several extensions, including framework enhancements, additional primitives, and relevant examples. The VisionWorks library of Computer Vision primitives with framework is optimized for NVIDIA platforms. VisionWorks can also be used together with other libraries such as OpenCV, providing access to additional computer vision algorithms or NPP, providing access to additional optimized performance primitives.

Applications can use the API provided by VisionWorks to build highly optimized computer vision functions by combining low-level primitives.

Components

The VisionWorks API consists of framework and data objects, vision primitives, and source samples.

  • Framework and data objects are objects such as graph and data containers needed for the execution of vision primitives. In addition to the OpenVX standard API (Objects and Advanced Objects), VisionWorks features a set of extra framework features (Framework APIs).
  • Primitives (also called vision functions or kernels) implement purely computational and stateless vision functions, that can be reused in many higher level vision pipelines. In addition to the OpenVX standard vision primitives (Vision Functions), VisionWorks features a number of extra primitives (Vision Primitives APIs).
  • Source samples demonstrate the use of the VisionWorks API with executable code examples.

The following diagram shows the overall VisionWorks toolkit architecture.

Note
Click the diagram below to jump to the specified module.
VisionWorks Architecture Diagram

Sample Applications VisionWorks API Extensions VisionWorks Framework Extensions OpenVX API Functions OpenVX Framework Objects CUDA Documentation NVIDIA Tegra Processor Default