Open ashtanmistal opened 3 weeks ago
Further hyperparameter tuning of the PointNet++ model could also, of course, reduce some of these incorrectly classified datapoints. Especially in the larger tree clusters that ought to not be classified as buildings.
Reduction of the buffer in the dataset preprocessing could also be useful, but that would only reduce the issue and not mitigate it.
The noise introduced will also somewhat be hidden by integration of Forest Friends, as a lot of the noise is caused by incorrectly classified tree datapoints in the PointNet++ model.
As a result further implementation and testing will have to be completed after Forest Friends is integrated.
LiDAR-DenseSeg semantic segmentation left some incorrectly classified data points in the resulting point cloud. Attempts were made to remove these datapoints and mitigate errors, but a more thorough process is required to better differentiate noise from sparse wall points.
What has been tried so far:
What is likely the solution (Will attempt after completion of Forest Friends) is the following:
Open3d can provide some of these functions; current library installation issues present that are less pressing than Forest Friends completion.