sloam
sloam copied to clipboard
training a new RangeNet++ model for different lidar
After substantial modifications to the code to receive and process our lidar clouds (32x2000) we realized the trained model provided, while working on your sample .bags
, could not identify tree trunks in our forestry dataset. So we need to train a new model using the RangeNet++ training like you did.
What configuration environment did you use? I'm trying 22.04 now. Will try 20.04.1 next with: tensorflow-2.7.0 Python 3.7-3.9 GCC 7.3.1 CuDNN 8.1 Cuda 11.2
The docker image and source installation instructions you provided for SLOAM were perfect.
Thanks for your time and attention.