features generated from Lift-Splat-Shot (LSS) [42] and our Sparse View Transformer. The pink area denotes non-empty voxels predicted from images.

1. Sparse View Transformer
   1. Image-Aware Sparse Module
   2. Depth-Aware Sparse Module
   3. how to implement it? [18]

SparseFusion: Efficient Sparse Multi-Modal Fusion Framework for Long-Range 3D Perception, 24

[18] BEVPoolv2: A cutting-edge implementation of bevdet toward deployment, 22