Point signature, a representation describing the structural neighborhood of a point in 3D shapes, can be
applied to establish correspondences between points in 3D shapes. Conventional methods apply a
weight-sharing network, e.g., graph neural network, across all neighborhoods to generate point signatures
directly and gain the generalization ability by extensive training over a large number of training samples
from scratch. However, these methods lack the flexibility in rapidly adapting to unseen neighborhood
structures and thus generalizes poorly on new point sets. In this paper, we propose a novel meta-learning
based 3D point signature model, named 3D meta point signature (MEPS) network,
that is capable of learning robust point signatures in 3D shapes. By regarding each point signature learning
process as a task, our method obtains an optimized model over the best performance on the distribution of
all tasks, generating reliable signatures for new tasks, i.e., signatures of unseen point neighborhoods.
Specifically, the MEPS consists of two modules: a base signature learner and a meta signature learner.
During training, the base-learner is trained to perform specific signature learning tasks. In the meantime,
the meta-learner is trained to update the base-learner with optimal parameters. During testing, the
meta-learner that is learned with the distribution of all tasks can adaptively change parameters of the
base-learner, accommodating to unseen local neighborhoods. We evaluate the MEPS model on a dataset for 3D
shape correspondence. Experimental results demonstrate that our method gains significant improvements over
the baseline model and achieves state-of-the-art results.
|
