Exploration of Energy-Efficient Architecture for Graph-Based Point-Cloud Deep Learning

摘要

Deep learning on point clouds has attracted increasing attention in the fields of 3D computer vision and robotics. In particular, graph-based point-cloud deep neural networks (DNNs) have demonstrated promising performance in 3D object classification and scene segmentation tasks. However, the scattered and irregular graph-structured data in a graph-based point-cloud DNN cannot be computed efficiently by existing SIMD architectures and accelerators. Following a review of the challenges of point-cloud DNN and the key edge convolution operation, we provide several directions in optimizing the processing architecture, including computation model, data reuse, and data locality, for achieving an effective acceleration and an improved energy efficiency.