Algorithm 2: Supervoxel-based projection
	Input:  Lower-view RGBD image ($I\in {[0,255]}^{H\times W\times 3}$ and $D\in {[0,255]}^{H\times W}$), upper-view 3D semantic reconstruction (including point cloud $P_{upper}\in {\mathbb{R}}^{N\times 3}$ and labels $L_{upper}\in {\{0,1,\cdots ,L\}}^N$, where $L\in \mathbb{N}$ denotes the number of semantic labels). Output:  Lower-view semantic segmentation ($S\in {\{0,1,\cdots ,L\}}^{H\times W}$)
1	Create a colored 3D point cloud ($P_{lower}$) from the lower-view RGB (I) and depth image (D).
2	Downsample the point cloud ($P_{lower}$) and keep track of each point’s original location.
3	Convert the lower-view point cloud ($P_{lower}$) to a voxel grid ($V_{lower}$).
4	Utilize SLIC on the voxel grid ($V_{lower}$), using masking to obtain supervoxels ($SV$).
5	Match the lower-view point cloud ($P_{lower}$) with the upper-view 3D semantic reconstruction ($P_{upper},L_{upper}$), determining the semantic label ($L_{lower}$) of each supervoxel ($SV$).
6	Project the semantic labels of the lower-view point cloud ($L_{lower}$) onto the image plane to obtain the semantic segmentation (S).