A Sketch Based 3D Shape Retrieval Approach Based on Efficient Deep Point-to-Subspace Metric Learning

TL;DR

This paper proposes a novel sketch-based 3D shape retrieval method utilizing efficient deep point-to-subspace metric learning to improve accuracy and retrieval speed.

Contribution

It introduces a new deep metric learning approach specifically designed for sketch-based 3D shape retrieval tasks.

Findings

Achieves higher retrieval accuracy compared to existing methods.

Reduces computational complexity in shape matching.

Demonstrates effectiveness on benchmark datasets.

Abstract

A sketch based 3D shape retrieval