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

Prajval Kumar Murali; Cong Wang; Ravinder Dahiya; Mohsen Kaboli

doi:10.1109/FLEPS53764.2022.9781490

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

Prajval Kumar Murali
, Cong Wang
, Ravinder Dahiya
, Mohsen Kaboli

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

4 Citations (Scopus)

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.

Original language	English
Title of host publication	2022 IEEE International Conference on Flexible and Printable Sensors and Systems, FLEPS 2022
Publisher	Institute of Electrical and Electronics Engineers
Number of pages	4
ISBN (Electronic)	978-1-6654-4273-2
DOIs	https://doi.org/10.1109/FLEPS53764.2022.9781490
Publication status	Published - 10 Jun 2022
Externally published	Yes
Event	2022 IEEE International Conference on Flexible and Printable Sensors and Systems (FLEPS) - Vienna, Austria Duration: 10 Jul 2022 → 13 Jul 2022

Conference

Conference	2022 IEEE International Conference on Flexible and Printable Sensors and Systems (FLEPS)
Country/Territory	Austria
City	Vienna
Period	10/07/22 → 13/07/22

Keywords

Point cloud compression
Temperature sensors
Temperature distribution
Three-dimensional displays
Costs
Annotations
Data collection

Access to Document

10.1109/FLEPS53764.2022.9781490

Cite this

@inproceedings{6b0a210eab214efa8382f64cf883d988,

title = "Towards Robust 3D Object Recognition with Dense-to-Sparse Deep Domain Adaptation",

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.",

keywords = "Point cloud compression, Temperature sensors, Temperature distribution, Three-dimensional displays, Costs, Annotations, Data collection",

author = "Murali, \{Prajval Kumar\} and Cong Wang and Ravinder Dahiya and Mohsen Kaboli",

year = "2022",

month = jun,

day = "10",

doi = "10.1109/FLEPS53764.2022.9781490",

language = "English",

booktitle = "2022 IEEE International Conference on Flexible and Printable Sensors and Systems, FLEPS 2022",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

note = "2022 IEEE International Conference on Flexible and Printable Sensors and Systems (FLEPS) ; Conference date: 10-07-2022 Through 13-07-2022",

}

Murali, PK, Wang, C, Dahiya, R & Kaboli, M 2022, Towards Robust 3D Object Recognition with Dense-to-Sparse Deep Domain Adaptation. in 2022 IEEE International Conference on Flexible and Printable Sensors and Systems, FLEPS 2022., 9781490, Institute of Electrical and Electronics Engineers, 2022 IEEE International Conference on Flexible and Printable Sensors and Systems (FLEPS), Vienna, Austria, 10/07/22. https://doi.org/10.1109/FLEPS53764.2022.9781490

Towards Robust 3D Object Recognition with Dense-to-Sparse Deep Domain Adaptation. / Murali, Prajval Kumar; Wang, Cong; Dahiya, Ravinder et al.
2022 IEEE International Conference on Flexible and Printable Sensors and Systems, FLEPS 2022. Institute of Electrical and Electronics Engineers, 2022. 9781490.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Towards Robust 3D Object Recognition with Dense-to-Sparse Deep Domain Adaptation

AU - Murali, Prajval Kumar

AU - Wang, Cong

AU - Dahiya, Ravinder

AU - Kaboli, Mohsen

PY - 2022/6/10

Y1 - 2022/6/10

N2 - 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.

AB - 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.

KW - Point cloud compression

KW - Temperature sensors

KW - Temperature distribution

KW - Three-dimensional displays

KW - Costs

KW - Annotations

KW - Data collection

U2 - 10.1109/FLEPS53764.2022.9781490

DO - 10.1109/FLEPS53764.2022.9781490

M3 - Conference contribution

BT - 2022 IEEE International Conference on Flexible and Printable Sensors and Systems, FLEPS 2022

PB - Institute of Electrical and Electronics Engineers

T2 - 2022 IEEE International Conference on Flexible and Printable Sensors and Systems (FLEPS)

Y2 - 10 July 2022 through 13 July 2022

ER -

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

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this