The preprocessing of input generation without comments is a bit difficult to understand, especially the choose and KNN indexing.
Do you know how to generate ground truth segmentation label and virtualize them and prediction for all points(pixels) in 3d for example via point cloud?
Hi Yisheng,
The preprocessing of input generation without comments is a bit difficult to understand, especially the choose and KNN indexing. Do you know how to generate ground truth segmentation label and virtualize them and prediction for all points(pixels) in 3d for example via point cloud?