Abstract. We introduce Spatial Group Convolution (SGC) for accelerating the computation of 3D dense prediction tasks. SGC is orthogonal to group convolution, which works on spatial dimensions rather than feature channel dimension. It divides input voxels into difffferent groups, then conducts 3D sparse convolution on these separated groups. As only valid voxels are considered when performing convolution, computation can be signifificantly reduced with a slight loss of accuracy. The proposed operations are validated on semantic scene completion task, which aims to predict a complete 3D volume with semantic labels from a single depth image. With SGC, we further present an effiffifficient 3D sparse convolutional network, which harnesses a multiscale architecture and a coarse-to-fifine prediction strategy. Evaluations are conducted on the SUNCG dataset, achieving state-of-the-art performance and fast speed