Video processing using gstreamer

Gisle Sælensminde

Half-day workshop - in English

Gstreamer is a pipeline-based open source media framework that let
you manipulate video in a programmatic fashion.

Say you have a machine learning system like face recognition or some
other computer vision algorithm working on images. Then you often want
to run the algorithm on the frames of a video rather than individual images.

This is often done by dumping the video frames to image files, and then
running your algorithm on the images. While this works in many cases,
it is slow and if you have real time requirements, like a number plate
or bar code reader, that won't work.

A faster and more reliable way is to use gstreamer and then make a
gstreamer filter to do the task. In this workshop you will learn to
set up a gstreamer video processing pipeline and build your own
gstreamer filter.

Primarily for: Developers

Participant requirements: A laptop with Linux or MacOS. Windows 7 and later with Visual studio can also be used, but expect that more time will be needed for setup in that case.