10-shot segmentation. We report pointwise IoU for 9 classes and the average IoU over all 16 classes of ShapeNet- Part. See text for details.

10-shot segmentation. We report pointwise IoU for 9 classes and the average IoU over all 16 classes of ShapeNet- Part. See text for details.

Source publication
Preprint
Full-text available
In this paper, we revisit the classical representation of 3D point clouds as linear shape models. Our key insight is to leverage deep learning to represent a collection of shapes as affine transformations of low-dimensional linear shape models. Each linear model is characterized by a shape prototype, a low-dimensional shape basis and two neural net...

Context in source publication

Context 1
... use the densely annotated ShapeNetPart [45] to evaluate the segmentation performance of our few-shot segmentation scheme. We report in Table 3 the performance of our 10-shot segmentation scheme for nine classes of ShapeNetCore, and the average performance over all 16 classes. As mentioned in Section 3.4, all the alignment and projection networks share a common PointNet [42] encoder which acts as a global feature extractor. ...