The first step to creating a docker image is to have the application tested in a working environment (e.g., ubuntu + gstreamer-xilinx1.0-pulseaudio + python3.10 and so on). It is possible to develop and test inside another base docker container (a sandbox running process of a docker image). This allows specific versions of dependencies that are needed for the application without disturbing the original workstations environment (e.g., having python 3.11 in the docker workspace versus 3.10 on a local working environment). A separate working environment is not mandatory, you can develop and test the application on a regular machine, provided it can install all the dependencies and set the required environment.
AMD also provides two example docker images used to rebuild example Docker containers.
kria-developerDocker image can be found on the Docker hub. It contains all the development libraries you need to build against.
kria-runtimeDocker image is a subset of the
kria-developerDocker image, and can also be found on the Docker hub. Because it only contains the runtime library dependencies, it is, therefore, a smaller footprint base image that can be used to install a pre-built app.