AttWalk: Attentive Cross-Walks for Deep Mesh Analysis
This addresses mesh analysis for 3D shape tasks, offering a novel approach to improve efficiency and accuracy, though it is incremental as it builds on existing random walk methods.
The paper tackled the problem of random walks in mesh representation potentially wandering into non-characteristic regions, which can harm shape analysis when few walks are used, by proposing a walk-attention mechanism that leverages mutual information between walks to extract a single descriptor, achieving state-of-the-art results in 3D shape classification and retrieval with only a handful of walks.
Mesh representation by random walks has been shown to benefit deep learning. Randomness is indeed a powerful concept. However, it comes with a price: some walks might wander around non-characteristic regions of the mesh, which might be harmful to shape analysis, especially when only a few walks are utilized. We propose a novel walk-attention mechanism that leverages the fact that multiple walks are used. The key idea is that the walks may provide each other with information regarding the meaningful (attentive) features of the mesh. We utilize this mutual information to extract a single descriptor of the mesh. This differs from common attention mechanisms that use attention to improve the representation of each individual descriptor. Our approach achieves SOTA results for two basic 3D shape analysis tasks: classification and retrieval. Even a handful of walks along a mesh suffice for learning.