Pillar-Based 3D Object Detection from Point Cloud with Multiattention Mechanism

Figure 1

Network overview. The network mainly consists of SOPA, a pillar feature network, SOCA, backbone network, SAPI, and SSD detection head.