Learning point embedding for 3D data processing
This addresses a limitation in point cloud processing for computer vision applications, offering a novel method to handle variable input sizes, though it is incremental in improving upon existing point-based approaches.
The paper tackles the problem of processing 3D point clouds with fixed-size constraints by proposing PE-Net, which learns point embeddings for high-dimensional representations, enabling standard 2D CNNs to handle unordered inputs and adapt to varying point numbers. It achieves state-of-the-art performance in classification and part segmentation on datasets like ModelNet and ShapeNetPart.
Among 2D convolutional networks on point clouds, point-based approaches consume point clouds of fixed size directly. By analysis of PointNet, a pioneer in introducing deep learning into point sets, we reveal that current point-based methods are essentially spatial relationship processing networks. In this paper, we take a different approach. Our architecture, named PE-Net, learns the representation of point clouds in high-dimensional space, and encodes the unordered input points to feature vectors, which standard 2D CNNs can be applied to. The recommended network can adapt to changes in the number of input points which is the limit of current methods. Experiments show that in the tasks of classification and part segmentation, PE-Net achieves the state-of-the-art performance in multiple challenging datasets, such as ModelNet and ShapeNetPart.