MortonNet: Self-Supervised Learning of Local Features in 3D Point Clouds
This addresses the need for effective self-supervised learning in 3D vision tasks, offering incremental improvements for point cloud segmentation.
The paper tackles the problem of learning local features in 3D point clouds without supervision, resulting in features that improve semantic segmentation performance by +3% on S3DIS and achieve a 3.8% improvement over state-of-the-art on vKITTI.
We present a self-supervised task on point clouds, in order to learn meaningful point-wise features that encode local structure around each point. Our self-supervised network, named MortonNet, operates directly on unstructured/unordered point clouds. Using a multi-layer RNN, MortonNet predicts the next point in a point sequence created by a popular and fast Space Filling Curve, the Morton-order curve. The final RNN state (coined Morton feature) is versatile and can be used in generic 3D tasks on point clouds. In fact, we show how Morton features can be used to significantly improve performance (+3% for 2 popular semantic segmentation algorithms) in the task of semantic segmentation of point clouds on the challenging and large-scale S3DIS dataset. We also show how MortonNet trained on S3DIS transfers well to another large-scale dataset, vKITTI, leading to an improvement over state-of-the-art of 3.8%. Finally, we use Morton features to train a much simpler and more stable model for part segmentation in ShapeNet. Our results show how our self-supervised task results in features that are useful for 3D segmentation tasks, and generalize well to other datasets.