-
Notifications
You must be signed in to change notification settings - Fork 116
Closed
Description
Dear Sir or Madam,
i am training a PointNet++ architecture now for many weeks. However training does not converge at all. The datset contains 100 unique pointclouds (pc_size: 2000 or 20.000) and with data augmentation i get additional 100 (total: 200). I tried it with 2-4 Labels and many different training parameters. I hope you can help me with a pretrained model, which i can then transfer for my task.
Thanks in advance.
Metadata
Metadata
Assignees
Labels
No labels