TL;DR of the paper, generated automatically.

The paper says its full-waveform LiDAR Transformer runs faster and more accurately.

• Speed: intensity-guided token pruning gives over 8x speedup, at real-time throughput.
• Accuracy: nearly 9-point F1 gain over prior state of the art on Ghost-FWL.
• Generalization: FWLAug keeps waveform structure, helps on unseen scenes.