About this document

Coolded by Bike Chen is a document available to read on EtoBox.

The document presents RangeViT, a vision transformer-based approach for 3D semantic segmentation of LiDAR point clouds in autonomous driving. It combines projection-based methods with ViTs, leveraging pre-trained models on large image datasets to improve segmentation performance while addressing the challenges of inductive bias and training data scarcity. The proposed method outperforms existing projection-based techniques on benchmark datasets such as nuScenes and SemanticKITTI.

Author

Bike Chen

Language

EN