Open bilalsattar opened 4 years ago
In the original dataset, there were labels for objects that were either very heavily occluded or completely hidden (for instance behind a wall). The filtering of objects with less than 10 lidar points was an attempt to remove those objects.
I haven’t tried the recent update of the dataset but this problem might have been fixed.
I didn't get why are you excluding label boxes with less than 10 lidar points?