Unsupervised discovery of parts, structure, and dynamics
Name
1903.05136.pdf
Description
Accepted version
Size
4.91 MB
Format
Adobe PDF
Checksum (MD5)
ad279bdb42e07d3271c75423a5a69b29
Author(s)
Xu, Zhenjia
Liu, Zhijian
Sun, Chen
Murphy, Kevin
Freeman, William T
Tenenbaum, Joshua B
Wu, Jiajun
Date Issued
May 2019
Journal
ICLR: International Conference on Learning Representations
Citation
Xu, Zhenjia et al. "Unsupervised discovery of parts, structure, and dynamics." ICLR 2019: 7th International Conference on Learning Representations, May 6-9, 2019, New Orleans, Louisiana: https://openreview.net/forum?id=rJe10iC5K7 ©2019 Author(s)
Version
Author's final manuscript
Abstract
Humans easily recognize object parts and their hierarchical structure by watching how they move; they can then predict how each part moves in the future. In this paper, we propose a novel formulation that simultaneously learns a hierarchical, disentangled object representation and a dynamics model for object parts from unlabeled videos. Our Parts, Structure, and Dynamics (PSD) model learns to, first, recognize the object parts via a layered image representation; second, predict hierarchy via a structural descriptor that composes low-level concepts into a hierarchical structure; and third, model the system dynamics by predicting the future. Experiments on multiple real and synthetic datasets demonstrate that our PSD model works well on all three tasks: segmenting object parts, building their hierarchical structure, and capturing their motion distributions.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://openreview.net/forum?id=rJe10iC5K7
