An MIL-Derived Transformer for Weakly Supervised Point Cloud Segmentation

Keyphrases

• Transformer 100%
• Unsupervised Learning 100%
• Multiple Instance Learning 100%
• Point Cloud Segmentation 100%
• Transformer Model 50%
• Source Code 25%
• Trainable 25%
• Point Cloud 25%
• Logit 25%
• Robustness to Noise 25%
• Art Performance 25%
• Rescale 25%
• Max Pooling 25%
• Model Optimization 25%
• Supervisory Signal 25%
• Class Activation Map 25%
• S3DIS 25%
• Weighted Pooling 25%
• Average Pooling 25%
• Equivariance 25%

Computer Science

• Multiple Instance 100%
• Instance Learning 100%
• Point Cloud Segmentation 100%
• Annotation 33%
• Regularization 33%
• Point Cloud 33%
• Art Performance 33%
• Average Pooling 33%
• Feature Map 33%