On learning the right attention point for feature enhancement

We present a novel attention-based mechanism to learn enhanced point features for point cloud processing tasks, e.g., classification and segmentation. Unlike prior studies, which were trained to optimize the weights of a pre-selected set of attention points, our approach learns to locate the best attention points to maximize the performance of a specific task, e.g., point cloud classification. Importantly, we advocate the use of single attention point to facilitate semantic understanding in point feature learning. Specifically, we formulate a new and simple convolution, which combines convolutional features from an input point and its corresponding learned attention point (LAP). Our attention mechanism can be easily incorporated into state-of-the-art point cloud classification and segmentation networks. Extensive experiments on common benchmarks, such as ModelNet40, ShapeNetPart, and S3DIS, all demonstrate that our LAP-enabled networks consistently outperform the respective original networks, as well as other competitive alternatives, which employ multiple attention points, either pre-selected or learned under our LAP framework.

Automated summary

We propose a new attention-based mechanism for learning enhanced point features in point cloud processing tasks. Unlike previous studies, our approach learns to locate the best attention points to optimize the performance of specific tasks. We advocate the use of single attention points for better semantic understanding in point feature learning.