Classification of occluded 2d objects using deep learning of 3d shape surfaces

Vasileios Tzitzilonis; George Apostolopoulos; Vassilios Kappatos; Evangelos Dermatas

doi:10.1145/3200947.3201061

Classification of occluded 2d objects using deep learning of 3d shape surfaces

Vasileios Tzitzilonis, George Apostolopoulos, Vassilios Kappatos, Evangelos Dermatas

University of Patras

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

This paper presents a novel deep learning method for partially occluded 2D object classification. A 2D Convolutional Neural Network (CNN) was trained with partial and whole images of the 3D models obtained from different camera views. The efficiency of the proposed method in classifying partial objects in 40 categories is more than 80% in most objects and exceeds 95% in some of them.

Original language	English
Title of host publication	Proceedings - 10th Hellenic Conference on Artificial Intelligence, SETN 2018
Publisher	Association for Computing Machinery / Special Interest Group on Programming Languages
Publication date	9. Jul 2018
ISBN (Electronic)	978-1-4503-6433-1
DOIs	https://doi.org/10.1145/3200947.3201061
Publication status	Published - 9. Jul 2018
Event	10th Hellenic Conference on Artificial Intelligence, SETN 2018 - Patras, Greece Duration: 9. Jul 2018 → 12. Jul 2018

Conference

Conference	10th Hellenic Conference on Artificial Intelligence, SETN 2018
Country/Territory	Greece
City	Patras
Period	09/07/2018 → 12/07/2018
Sponsor	Hellenic Artificial Intelligence Society (EETN), University of Patras, University of Thessaly

Keywords

3D-Models
Classification
Computer Vision
Convolutional Neural Networks
Deep Learning
Machine learning

Documents & Links

10.1145/3200947.3201061

Cite this

@inproceedings{a62eed9e2dc248688c9ed57ba5cbb244,

title = "Classification of occluded 2d objects using deep learning of 3d shape surfaces",

abstract = "This paper presents a novel deep learning method for partially occluded 2D object classification. A 2D Convolutional Neural Network (CNN) was trained with partial and whole images of the 3D models obtained from different camera views. The efficiency of the proposed method in classifying partial objects in 40 categories is more than 80% in most objects and exceeds 95% in some of them.",

keywords = "3D-Models, Classification, Computer Vision, Convolutional Neural Networks, Deep Learning, Machine learning",

author = "Vasileios Tzitzilonis and George Apostolopoulos and Vassilios Kappatos and Evangelos Dermatas",

year = "2018",

month = jul,

day = "9",

doi = "10.1145/3200947.3201061",

language = "English",

booktitle = "Proceedings - 10th Hellenic Conference on Artificial Intelligence, SETN 2018",

publisher = "Association for Computing Machinery / Special Interest Group on Programming Languages",

note = "10th Hellenic Conference on Artificial Intelligence, SETN 2018 ; Conference date: 09-07-2018 Through 12-07-2018",

}

Tzitzilonis, V, Apostolopoulos, G, Kappatos, V & Dermatas, E 2018, Classification of occluded 2d objects using deep learning of 3d shape surfaces. in Proceedings - 10th Hellenic Conference on Artificial Intelligence, SETN 2018. Association for Computing Machinery / Special Interest Group on Programming Languages, 10th Hellenic Conference on Artificial Intelligence, SETN 2018, Patras, Greece, 09/07/2018. https://doi.org/10.1145/3200947.3201061

Classification of occluded 2d objects using deep learning of 3d shape surfaces. / Tzitzilonis, Vasileios; Apostolopoulos, George; Kappatos, Vassilios et al.
Proceedings - 10th Hellenic Conference on Artificial Intelligence, SETN 2018. Association for Computing Machinery / Special Interest Group on Programming Languages, 2018.

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - Classification of occluded 2d objects using deep learning of 3d shape surfaces

AU - Tzitzilonis, Vasileios

AU - Apostolopoulos, George

AU - Kappatos, Vassilios

AU - Dermatas, Evangelos

PY - 2018/7/9

Y1 - 2018/7/9

N2 - This paper presents a novel deep learning method for partially occluded 2D object classification. A 2D Convolutional Neural Network (CNN) was trained with partial and whole images of the 3D models obtained from different camera views. The efficiency of the proposed method in classifying partial objects in 40 categories is more than 80% in most objects and exceeds 95% in some of them.

AB - This paper presents a novel deep learning method for partially occluded 2D object classification. A 2D Convolutional Neural Network (CNN) was trained with partial and whole images of the 3D models obtained from different camera views. The efficiency of the proposed method in classifying partial objects in 40 categories is more than 80% in most objects and exceeds 95% in some of them.

KW - 3D-Models

KW - Classification

KW - Computer Vision

KW - Convolutional Neural Networks

KW - Deep Learning

KW - Machine learning

U2 - 10.1145/3200947.3201061

DO - 10.1145/3200947.3201061

M3 - Article in proceedings

AN - SCOPUS:85052012912

BT - Proceedings - 10th Hellenic Conference on Artificial Intelligence, SETN 2018

PB - Association for Computing Machinery / Special Interest Group on Programming Languages

T2 - 10th Hellenic Conference on Artificial Intelligence, SETN 2018

Y2 - 9 July 2018 through 12 July 2018

ER -

Classification of occluded 2d objects using deep learning of 3d shape surfaces

Abstract

Conference

Keywords

Documents & Links

Fingerprint

Cite this