again, I hope the category is ok for Mona code, otherwise I’d gladly post somewhere else.
I was able to get the minimal docker version of mona installed and to detect vehicles on reference_drives/raw/east_outgoing/east-000-000.yaml though if I try to do post processing on output/east-000-000 I get an error message that seems to indicate two issues. One is that an EstimatorActor object has no attribute deserializer, another is that it seems in an init function, it is tried to open a file from a local cache and read a number from it, however that file does not contain a number so some arbitrary string is read. I post the full error here, it’s only that I investigated by printing out both variables of that subtraction operation and found how one of them is not a number but something weird.
Hi,
I suspect that you didn’t have Git LFS installed while creating the docker image. Therefore, the terrain model file would be only a pointer file and does not contain the actual data. To confirm, please check the content of mona/dgm/69_533t50dgm.txt · release · tum-cps / MONA Dataset · GitLab in your local working tree. Does it contain many lines of three tab separated numbers? If not, please re-checkout the repository with Git LFS installed and recreate the docker image.
Hi,
one remark that I might want to add regarding the terrain model:
For the creation of the full dataset, we have used a terrain model with a grid size of 1m. However, this model can not be freely distributed due to license restrictions. The provided model has a grid size of 50m and is free, but due to the coarser resolution the results may differ to some extent.
thank you for your support that was exactly the problem! Sorry it failed without error message and I run docker in a virtual machine. When I reinstalled Github on that VM I forgot to add the LFS support. I was able to run the post processing without any further issues.