Notebook 01: PointNet and PointNet++ baselines on ForestSemantic-MS

Status
done
Question
How well do plain PointNet and PointNet++ segment low vegetation at 4,096 points per 5 m block?

Random sampling, xyz plus height above ground, plain cross-entropy, 40 epochs, with train plot 4 used for validation.

Results

Results are not public yet.

Related

Papers: First paper · height sampling

Datasets: ForestSemantic-MS

Methods and baselines: PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space