Towards Robust 3D Object Recognition with Dense-to-Sparse Deep Domain Adaptation

TL;DR

This paper introduces a novel unsupervised domain adaptation method that improves 3D object recognition in sparse point clouds by leveraging dense point cloud training, reducing data collection and annotation costs.

Contribution

The work presents a new approach for dense-to-sparse domain adaptation in 3D object recognition, achieving competitive results without training on sparse data.

Findings

Competitive performance on both dense and sparse point clouds

Effective domain adaptation from dense to sparse data

Reduced need for sparse data annotation and retraining

Abstract

Three-dimensional (3D) object recognition is crucial for intelligent autonomous agents such as autonomous vehicles and robots alike to operate effectively in unstructured environments. Most state-of-art approaches rely on relatively dense point clouds and performance drops significantly for sparse point clouds. Unsupervised domain adaption allows to minimise the discrepancy between dense and sparse point clouds with minimal unlabelled sparse point clouds, thereby saving additional sparse data collection, annotation and retraining costs. In this work, we propose a novel method for point cloud based object recognition with competitive performance with state-of-art methods on dense and sparse point clouds while being trained only with dense point clouds.