A SAE stage that maps object locations from camera / pixel space to geo-coordinate space. For that it needs to be configured with a number of optical and geometrical camera parameters.
Due to cameratransform
being stuck on Python 3.11 (due to Pandas 1.x), we need to use pyenv for this repository.
- Make sure
pyenv
is installed - Install the latest Python 3.11 (e.g. Python 3.11.9 at the time of writing):
pyenv install 3.11.9
- If you encounter installation error, check that you have all Python build dependencies installed (refer to pyenv README)
- Set the installed Python version for this directory, e.g.
pyenv local 3.11.9
- Run
poetry install
, it should automatically pick up the correct Python version and use it for the virtualenv
- Input message must be a
SaeMessage
. The geo-mapping is done on eachDetection
message within. If there are noDetection
messages, the processing is effectively a no-op. - Output is the input
SaeMessage
with geo-coordinates added to everyDetection
. All other fields are preserved.