Prototype-based multi-view fine-grained 3D classification and ad-hoc interpretability

doi:10.1016/j.patcog.2026.113596

Prototype-based multi-view fine-grained 3D classification and ad-hoc interpretability

2026-03-26

0

PRE

AI

DOI：10.1016/j.patcog.2026.113596

Original

Original request for help

Save

Abstract

En 中文

• A novel prototype learning framework is proposed for fine-grained 3D classification. • Ad-hoc interpretability enables case-based reasoning through global and local views. • Proto-FG3D outperforms SOTA models in accuracy and efficiency on FG3D and ModelNet40.

Keywords:

Prototype learning

Fine-grained classification

3D classification

Multi-view

Interpretability

Prototype-based multi-view fine-grained 3D classification and ad-hoc interpretability

Abstract

Journal

Organization