Semantic Masked Autoencoder uses learned component prototypes to mask complete point cloud parts during pre-training and as prompts during fine-tuning, improving downstream 3D classification and segmentation.
Con- trast with reconstruct: Contrastive 3d representation learn- ing guided by generative pretraining
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Exploring Semantic Masked Autoencoder for Self-supervised Point Cloud Understanding
Semantic Masked Autoencoder uses learned component prototypes to mask complete point cloud parts during pre-training and as prompts during fine-tuning, improving downstream 3D classification and segmentation.