Napredna pretraga

Pregled bibliografske jedinice broj: 1028057

On using PointNet Architecture for Human Body Segmentation


Jertec, Andrej; Bojanić, David; Bartol, Kristijan; Pribanić, Tomislav; Petković, Tomislav; Petrak , Slavenka
On using PointNet Architecture for Human Body Segmentation // 2019 11th International Symposium on Image and Signal Processing and Analysis (ISPA) / Lončarić, Sven ; Bregović, Robert ; Carli, Marco ; Subašić, Marko (ur.).
Dubrovnik: IEEE, 2019. str. 253-257 doi:10.1109/ISPA.2019.8868844 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), ostalo)


Naslov
On using PointNet Architecture for Human Body Segmentation

Autori
Jertec, Andrej ; Bojanić, David ; Bartol, Kristijan ; Pribanić, Tomislav ; Petković, Tomislav ; Petrak , Slavenka

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), ostalo

Izvornik
2019 11th International Symposium on Image and Signal Processing and Analysis (ISPA) / Lončarić, Sven ; Bregović, Robert ; Carli, Marco ; Subašić, Marko - Dubrovnik : IEEE, 2019, 253-257

ISBN
978-1-7281-3140-5

Skup
11th International Symposium on Image and Signal Processing and Analysis

Mjesto i datum
Dubrovnik, Hrvatska, 23-25.09.2019

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
PointNet, human body segmentation, 3D shape analysis, deep learning

Sažetak
In the case of structured data, such as 2D images, many variants of traditional convolution neural network architectures have been successfully proposed. Learning from unstructured sets of data, such as sets of 3D point clouds, is a challenging task due to numerous reasons among which two most important ones are: 3D point cloud is generally (i) unordered and (ii) sparse data set. Therefore, the architectures have been proposed which are invariant to both ordering and number of points in the point cloud. PointNet is one such architecture, originally introduced and demonstrated on the task of classification and segmentation of the ModelNet40 data set. In this work we study the performance of PointNet on an even more demanding task, segmentation of human body parts. Finding enough training data of enough quality is generally a problem in deep learning, and especially for human body segmentation. To that end we take advantage of SMPL model which provides human body models in many shapes and sizes in an essentially automatic fashion, therefore avoiding a cumbersome procedure of manual collection and preparation of training data. Our results show that the proposed PointNet variant trained using SMPL model provides competitive segmentation results on the task of human body segmentation.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo, Interdisciplinarne tehničke znanosti



POVEZANOST RADA


Projekt / tema
HRZZ-IP-2018-01-8118 - Izračun antropometrijskih mjera pametnim telefonom i tabletom (Tomislav Pribanić, )
KK.01.1.1.01.0009 - Napredne metode i tehnologije u znanosti o podatcima i kooperativnim sustavima (Sven Lončarić, EK)

Ustanove
Fakultet elektrotehnike i računarstva, Zagreb,
Tekstilno-tehnološki fakultet, Zagreb

Citati