embedded Vision SDK

embedded Vision SDK

Embedded Computer Vision Software Development Kit

Au-Zone’s embedded Computer Vision (eCV) SDK provides developers with the necessary tools and resources to dramatically increase performance of advanced computer vision algorithms on GPU enabled System on Chip (SoC) processors.  An immediate performance improvement of 5 - 20X can be expected for many common OpenCV functions as compared to execution on the CPU alone.  

In addition to the performance gains achieved by accelerating vision algorithms, developers also significantly lower engineering cost, time to revenue and technical risk by using the pre-tested and fully optimized solution.

Have questions?  Just send us a quick email


Realtime Object Tracking using AMF for background subtraction on i.MX6 GPU

Each tile represents a sequential stage in the pipeline, starting with input on top left.   Processing time for each stage in the algorithm is also shown and system easily performs all functions at 30 frames per second


Realtime 720p Sobel Filter on i.MX6 GPU

Demonstration of a 2 dimensional Sobel operator used for edge detection.  In this demo, the untouched 720p source video is shown on the left with the filtered output on the right.

eCV SDK Product Brief

eCV SDK Datasheet (registration required)

The SDK is designed specifically to help developers create and accelerate new vision algorithms for embedded platforms as well as to port existing OpenCV based algorithms to deeply embedded hardware platforms.  In addition to the core vision acceleration features provided by the eCV Vision Library module, the SDK also delivers complete, out of the box support for many common smart camera functions.  This includes support for several popular image sensors, fully hardware accelerated vision pipeline, optimized I/O management and scalable graphical interface for overlays and GUIs. 

The SDK also includes an integrated development environment (IDE), quick start guide, tutorials and a complete Linux BSP to enable fast prototyping of embedded vision solutions on Freescale's i.MX6 family of processors. 

Whether you’re starting your algorithm development from a blank sheet, a tested Matlab model or an existing OpenCV implementation, the eCV SDK will help you migrate and optimize your implementation on a custom embedded hardware platform.

Standard Features:

  • Desktop IDE and emulation environment
  • Algorithm creation, instrumentation & optimization
  • Build & debug tools for embedded target
  • Library of accelerated low level CV primitives
  • Performance Analyzer for embedded target


Advanced Features:


  • Background subtraction
  • Blob detection and tracking
  • Object detection & classification
  • AMF, Optical Flow, DIFF, MOG2, HOG, SVM
  • Multi camera surround view



  • Colour space conversions
  • Lens Distortion Correction (LDC)
  • Image optimization & transforms
  • Arithmetic functions
  • Filters (Sobel, blur, dilate etc)
  • Image morphology (top hat, bottom hat, open, close)



  • OpenCV 2.4
  • OpenGL|ES 3.0
  • OpenVG and OpenCL
  • GStreamer Plugin Framework
  • Qt Framework Integration
  • Eigen (C++ library)
  • Robot Operating System (ROS) Interface



  • SVM (Support Vector Machine)
  • Cascade Classifiers



  • Vector Graphics for UI primitives OSD and bounding boxes
  • Qt integration  with support for h/w accelerated, rich UI and integrated video rendering


Typical Applications:

  • Business Intelligence - intelligent monitoring and data collection
  • Activity monitoring - worksite safety, healthcare
  • Driver / Operator Assist System for heavy equipment and commercial vehicles
  • Building Automation - occupancy detection, count, flow
  • Anonymous Audience Analytics - demographics & engagement
  • Intelligent Traffic Systems - vehicle count, volume, velocity, class
  • Autonomous vehicles and robotics
  • Access, Security and Surveillance

Contact us today for a free 30 minute consultation to discuss how Au-Zone's Design Services and Imaging Solutions can help you save time and money on your next project.