Rerun is a great tool for realtime data visualization. You simply start Rerun viewer, and then you can log many types of data from anywhere in your python script like so:
import rerun
rerun.init("my_project") # required before logging
rerun.log_image("zed_top", image)
rerun.log_scalar("force_z", force[2])
...See the example notebook for more.
ℹ️ A note on starting the Rerun viewer: you can start it by calling
rerun.spawn()from Python. However when starting Rerun like that, there is no way to specify a memory limit. This quickly becomes a problem when logging images, so we recommend starting Rerun from a terminal:python -m rerun --memory-limit 16GB
Some limitations to be aware of:
- Rerun is for one-way communication. You cannot get user input, such of confirmation of clicks from Rerun back to your Pythn scripts. If you want that, you can create OpenCV windows.
- Rerun is best for non-persistent data.
You can record everything sent to Rerun in a
.rrdfile, however it's not clear how easily get your data out of that format. So you probably want to log some images or videos of your experiment to disk youself.
We provide a MultiprocessRerunRGBLogger class that can be used to log a camera feed to Rerun when using a camera through the MultiprocessRGBPublisher class.
In its own process it simply checks the shared memory created by the publisher, and when it detects a new image, it logs it to Rerun.