I am currently working on strengthening a product with 3 jetsons TX2, i.e. making it production ready.
- Part of this process goes through CI/CD, which I create a topic for: https://devtalk.nvidia.com/default/topic/1045237/how-would-you-set-up-a-ci-cd-continuous-integration-deployment-on-a-jetson-tx2/#5303553
- Part of this process is setting up the dev and prod environments as similar as possible, which is the subject of this topic.
For this second purpose, I am trying to use python virtual environments on the Jetson.
I would like to be able to choose what version of python my application is running with, with the appropriate versions of openCV and tensorflow.
There are some articles that point how to achieve this with OpenCV, e.g.:
- How to Install OpenCV (3.4.0) on Jetson TX2: https://jkjung-avt.github.io/opencv3-on-tx2/
But I have the feeling that I am missing the big picture.
I am therefore looking for help: “how would you have multiple isolated python environments on your jetson, each of them being able to make use of the appropriate version of openCV and tensorflow (and without breaking each other) ?”