Linked Dynamic Graph CNN: Learning on Point Cloud via Linking Hierarchical Features
This addresses the challenge of robust environmental understanding for robots using point cloud data, representing an incremental improvement over existing graph CNN methods.
The paper tackles the problem of learning on point clouds, which are sparse and unstructured, by proposing a linked dynamic graph CNN (LDGCNN) for classification and segmentation, achieving state-of-the-art performance on ModelNet40 and ShapeNet datasets.
Learning on point cloud is eagerly in demand because the point cloud is a common type of geometric data and can aid robots to understand environments robustly. However, the point cloud is sparse, unstructured, and unordered, which cannot be recognized accurately by a traditional convolutional neural network (CNN) nor a recurrent neural network (RNN). Fortunately, a graph convolutional neural network (Graph CNN) can process sparse and unordered data. Hence, we propose a linked dynamic graph CNN (LDGCNN) to classify and segment point cloud directly in this paper. We remove the transformation network, link hierarchical features from dynamic graphs, freeze feature extractor, and retrain the classifier to increase the performance of LDGCNN. We explain our network using theoretical analysis and visualization. Through experiments, we show that the proposed LDGCNN achieves state-of-art performance on two standard datasets: ModelNet40 and ShapeNet.