Open access
Sep 2026
GSSP-KAN: An Efficient Kansformer-Based Network with Grouped Separable Sparse Convolution for Large-Scale LiDAR Point Cloud Semantic Segmentation
The proposed GSSP-KAN is an efficient semantic segmentation network that integrates Kansformer with grouped separable sparse convolution, while the Grouped Separable Sparse Convolution Block (GSSP_Block) reduces redundant self-attention computation and preserves fine-grained local geometric structures.
F. Shuang, Lin-Yang Zou, Yong Li et al.
· Remote Sensing · 0 citations