Geodesic-based 2D size function and neural classifier for 3D object recognition

doi:10.1016/j.eswa.2026.132269

Geodesic-based 2D size function and neural classifier for 3D object recognition

2026-04-02

0

PRE

AI

M

Mohammed Ayoub Alaoui Mhamdi *

D

Djemel Ziou

DOI：10.1016/j.eswa.2026.132269

Original

Original request for help

Save

Abstract

En 中文

• Proposed a novel 2D size function combining geodesic eccentricity and displacement • Demonstrated invariance to affine transformations, articulations, and torsions • Integrated metric learning via neural networks for robust 3D object comparison • Validated on McGill, SHREC¡¦15, and Deformable Partial Shape Matching datasets • Achieved state-of-the-art accuracy in classifying articulated and incomplete 3D shapes

Keywords:

Geodesic eccentricity

2D size function

neural classifier

3D object recognition

metric learning

Geodesic-based 2D size function and neural classifier for 3D object recognition

Abstract

Journal

Organization