Recommendations for accuracy improvements
Posted: Tue Nov 26, 2024 8:10 am
I am working with airborne LiDAR datasets. As is often the case, the class distribution is heavily skewed. Some classes are much more prevalent than others.
Any recommendations for dealing with this?
One of my thoughts was to implementing a data augmentation workflow, but as I think we can only use one CTX cloud, it is not a great idea.
Also, there is no way to adjust the weighting for different classes, I think?
Any recommendations for dealing with this?
One of my thoughts was to implementing a data augmentation workflow, but as I think we can only use one CTX cloud, it is not a great idea.
Also, there is no way to adjust the weighting for different classes, I think?