By masking high-attention patches and aligning the features of masked and unmasked views, PointACL improves point cloud classification, part segmentation, and few-shot learning over baseline pretraining methods.
Contrast with reconstruct: Contrastive 3d representation learning guided by generative pretraining
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Point Cloud Understanding via Attention-Driven Contrastive Learning
By masking high-attention patches and aligning the features of masked and unmasked views, PointACL improves point cloud classification, part segmentation, and few-shot learning over baseline pretraining methods.