point-cloud-augmentation

PatchAugment: Local Neighborhood Augmentation in Point Cloud Classification

Data augmentation has become an essential technique in training deep neural network models to overcome limitations such as overfitting, reduced robustness, and lower generalization. Methods using 3D datasets are among the most common to use data augmentation techniques. However, these techniques are often applied to the entire object, ignoring the object’s local geometry. This is where PatchAugment comes in. What is PatchAugment? PatchAugment is a data augmentation framework that applies diff

PointAugment

PointAugment is an innovative auto-augmentation framework that can enrich the data diversity for classification networks when we train them. It uses a sample-aware approach and an adversarial learning strategy to optimize an augmentor network and a classifier network together. This way, the augmentor network can learn to produce modified samples that best fit the classifier network. Auto-Augmentation Framework for Classification Networks PointAugment is designed to enhance the quality of poin