From Point Cloud Perception Toward People Detection

Assia Belbachir, Assia Belbachir, Antonio Ortiz, Atle Aalerud, Ahmed Belbachir

2023

Abstract

Point clouds have become significant data inputs for 3D representation, enabling accurate analysis of 3D scenes and objects. People detection from point clouds is a challenging task due to data sparsity, irregularity, occlusion, and real-time detection constraints. Existing methods based on handcrafted features or deep learning have limitations in handling occlusions, pose variations, and fast detection. This paper introduces a Random Forest classifier for people detection in point clouds, aiming to achieve both accuracy and fast performance. The point cloud data are acquired using a multi-point LiDAR system. First experiments demonstrate the effectiveness of the approach and its efficient detection compared to Multiple Layer Perceptron (MLP) in our collected Dataset.

Paper Citation

in Harvard Style

Belbachir A., Ortiz A., Aalerud A. and Belbachir A. (2023). From Point Cloud Perception Toward People Detection. In Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO; ISBN 978-989-758-670-5, SciTePress, pages 520-526. DOI: 10.5220/0012258800003543

in Bibtex Style

@conference{icinco23,
author={Assia Belbachir and Antonio Ortiz and Atle Aalerud and Ahmed Belbachir},
title={From Point Cloud Perception Toward People Detection},
booktitle={Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO},
year={2023},
pages={520-526},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012258800003543},
isbn={978-989-758-670-5},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO
TI - From Point Cloud Perception Toward People Detection
SN - 978-989-758-670-5
AU - Belbachir A.
AU - Ortiz A.
AU - Aalerud A.
AU - Belbachir A.
PY - 2023
SP - 520
EP - 526
DO - 10.5220/0012258800003543
PB - SciTePress