Fig. 5: Experimental semantic segmentation on S3DIS dataset.
From: Random memristor-based dynamic graph CNN for efficient point cloud learning at the edge
a Schematic of RDGCNN for the semantic segmentation task on S3DIS dataset. b Comparison of semantic segmentation results of each test area in a 6-fold cross-validation between software trainable baseline and our co-design. c, Parameter comparison of trainable software baseline and our co-design. d, Representative semantic segmentation results of our co-design. e, Comparison of training costs between the fully trainable DGCNN, DGCNN with splitted EdgeConv, and RDGCNN with splitted random EdgeConv (ours). f, Comparison of inference energy between GPU, NPU, and our co-design.
