APoX-M: Accelerate deep point cloud analysis via adaptive graph construction

Lei Dai, Shengwen Liang, Ying Wang, Huawei Li, Xiaowei Li

Abstract: Graph-based deep learning point cloud processing has gained increasing popularity but its performance is dragged by the dominating graph construction (GC) phase with irregular computation and memory access. Existing works that accelerate GC by tailoring architecture for a single GC algorithm fail to maintain efficiency because they neglect the best GC algorithm variation incurred by the point-cloud density variation in changing scenarios. Therefore, we propose APoX-M, a unified architecture with an adaptive GC scheme that can identify the optimum GC approach according to the point cloud variation. We also provide better memory management and scheduling optimizations for better performance. Experiments indicate that APoX-M achieves higher performance and energy efficiency over existing accelerators.