Search papers, labs, and topics across Lattice.
This paper introduces a hyper-network architecture that predicts non-linear neural functional maps (NFM) to enhance the alignment of spectral bases in 3D shape matching. By employing a multi-layer perceptron (MLP) with skip-connections and conditioning the hyper-network on standard functional maps, the method addresses challenges such as partiality and topological noise. Experimental results show significant improvements in matching accuracy when integrated into existing unsupervised deep functional map pipelines, highlighting the effectiveness of the proposed approach in complex scenarios.
Non-linear neural functional maps can drastically improve 3D shape matching accuracy in the presence of noise and partial data.
Functional maps are the cornerstone of recent non-rigid 3D shape matching methods due to their efficiency and performance. However, existing methods struggle with challenging scenarios, such as partiality, topological noise, and raw point clouds. A primary bottleneck is that significant intrinsic distortion prevents truncated spectral bases from being accurately aligned via linear transformations (i.e., functional maps). To address this, we introduce a hyper-network that predicts non-linear neural functional maps (NFM), learned in an unsupervised manner, to better align spectral bases. Specifically, we model the NFM as an MLP with skip-connection to refine standard FM and employ a hyper-network to predict its weights, conditioned on standard FM. Our framework is trained using a novel unsupervised spectral alignment loss. Experiments demonstrate that our approach can be seamlessly integrated into state-of-the-art unsupervised deep functional map pipelines, substantially improving matching accuracy in demanding scenarios.
