Learning Rotation-Invariant Representations of Point Clouds Using Aligned Edge Convolutional Neural Networks
This work addresses the challenge of rotation variance in point cloud analysis for researchers and practitioners using deep learning, offering a more robust and efficient solution.
This paper introduces the Aligned Edge Convolutional Neural Network (AECNN) to learn rotation-invariant feature representations of point clouds. It achieves this by aligning local features with respect to automatically computed Local Reference Frames (LRFs), outperforming state-of-the-art methods in rotation robustness on classification and part segmentation tasks without data augmentation.
Point cloud analysis is an area of increasing interest due to the development of 3D sensors that are able to rapidly measure the depth of scenes accurately. Unfortunately, applying deep learning techniques to perform point cloud analysis is non-trivial due to the inability of these methods to generalize to unseen rotations. To address this limitation, one usually has to augment the training data, which can lead to extra computation and require larger model complexity. This paper proposes a new neural network called the Aligned Edge Convolutional Neural Network (AECNN) that learns a feature representation of point clouds relative to Local Reference Frames (LRFs) to ensure invariance to rotation. In particular, features are learned locally and aligned with respect to the LRF of an automatically computed reference point. The proposed approach is evaluated on point cloud classification and part segmentation tasks. This paper illustrates that the proposed technique outperforms a variety of state of the art approaches (even those trained on augmented datasets) in terms of robustness to rotation without requiring any additional data augmentation.