PointWOLF
PointWOLF copied to clipboard
Official Implementation (PyTorch) of "Point Cloud Augmentation with Weighted Local Transformations", ICCV 2021
how to fix this
Thank you for your work. Will you intend to release your code for training and testing on ShapeNet? Just as in Table 3 in your paper. Thanks!
Hi! Thanks for the awesome work! Where can we access the code to train PointWOLF for SceneObjectNN? Best.
DGCNN code ends with F.log_softmax but PointNet code does not, this results incorrectly in negative loss and poor training.