Li Yi, Leonidas J. Guibas, Aaron Hertzmann, Vladimir G. Kim, Hao Su, Ersin Yumer: Learning Hierarchical Shape Segmentation and Labeling from Online Repositories. ACM Trans. Graph. 36(4): 70:1-70:12 (2017)

Abstract:

We propose a method for converting geometric shapes into hierar- chically segmented parts with part labels. Our key idea is to train category-specific models from the scene graphs and part names that accompany 3D shapes in public repositories. These freely- available annotations represent an enormous, untapped source of information on geometry. However, because the models and cor- responding scene graphs are created by a wide range of modelers with different levels of expertise, modeling tools, and objectives, these models have very inconsistent segmentations and hierarchies with sparse and noisy textual tags. Our method involves two anal- ysis steps. First, we perform a joint optimization to simultaneously cluster and label parts in the database while also inferring a canon- ical tag dictionary and part hierarchy. We then use this labeled data to train a method for hierarchical segmentation and labeling of new 3D shapes. We demonstrate that our method can mine complex information, detecting hierarchies in man-made objects and their constituent parts, obtaining finer scale details than existing alterna- tives. We also show that, by performing domain transfer using a few supervised examples, our technique outperforms fully-supervised techniques that require hundreds of manually-labeled models.

Bibtex: 

@article{yghksy-lhsslor-17,
  title={Learning hierarchical shape segmentation and labeling from online repositories},
  author={Yi, Li and Guibas, Leonidas and Hertzmann, Aaron and Kim, Vladimir G and Su, Hao and Yumer, Ersin},
  journal={arXiv preprint arXiv:1705.01661},
  year={2017}
}