Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

View-Invariant Action Recognition From RGB Data via 3D Pose Estimation

Abstract: 

In this paper, we propose a novel view-invariant action recognition method using a single monocular RGB camera. View-invariance remains a very challenging topic in 2D action recognition due to the lack of 3D information in RGB images. Most successful approaches make use of the concept of knowledge transfer by projecting 3D synthetic data to multiple viewpoints. Instead of relying on knowledge transfer, we propose to augment the RGB data by a third dimension by means of 3D skeleton estimation from 2D images using a CNN-based pose estimator. In order to ensure view-invariance, a pre-processing for alignment is applied followed by data expansion as a way for denoising. Finally, a Long-Short Term Memory (LSTM) architecture is used to model the temporal dependency between skeletons. The proposed network is trained to directly recognize actions from aligned 3D skeletons. The experiments performed on the challenging Northwestern-UCLA dataset show the superiority of our approach as compared to state-of-the-art ones.

up
0 users have voted:

Paper Details

Authors:
Enjie Ghorbel, Konstantinos Papadopoulos, Girum G. Demisse, Djamila Aouada, Björn Ottersten
Submitted On:
8 May 2019 - 7:19am
Short Link:
Type:
Poster
Event:
Presenter's Name:
Renato Baptista
Document Year:
2019
Cite

Document Files

ICASSP_Renato_final.pdf

(36)

Subscribe

[1] Enjie Ghorbel, Konstantinos Papadopoulos, Girum G. Demisse, Djamila Aouada, Björn Ottersten, "View-Invariant Action Recognition From RGB Data via 3D Pose Estimation", IEEE SigPort, 2019. [Online]. Available: http://sigport.org/4073. Accessed: Dec. 08, 2019.
@article{4073-19,
url = {http://sigport.org/4073},
author = {Enjie Ghorbel; Konstantinos Papadopoulos; Girum G. Demisse; Djamila Aouada; Björn Ottersten },
publisher = {IEEE SigPort},
title = {View-Invariant Action Recognition From RGB Data via 3D Pose Estimation},
year = {2019} }
TY - EJOUR
T1 - View-Invariant Action Recognition From RGB Data via 3D Pose Estimation
AU - Enjie Ghorbel; Konstantinos Papadopoulos; Girum G. Demisse; Djamila Aouada; Björn Ottersten
PY - 2019
PB - IEEE SigPort
UR - http://sigport.org/4073
ER -
Enjie Ghorbel, Konstantinos Papadopoulos, Girum G. Demisse, Djamila Aouada, Björn Ottersten. (2019). View-Invariant Action Recognition From RGB Data via 3D Pose Estimation. IEEE SigPort. http://sigport.org/4073
Enjie Ghorbel, Konstantinos Papadopoulos, Girum G. Demisse, Djamila Aouada, Björn Ottersten, 2019. View-Invariant Action Recognition From RGB Data via 3D Pose Estimation. Available at: http://sigport.org/4073.
Enjie Ghorbel, Konstantinos Papadopoulos, Girum G. Demisse, Djamila Aouada, Björn Ottersten. (2019). "View-Invariant Action Recognition From RGB Data via 3D Pose Estimation." Web.
1. Enjie Ghorbel, Konstantinos Papadopoulos, Girum G. Demisse, Djamila Aouada, Björn Ottersten. View-Invariant Action Recognition From RGB Data via 3D Pose Estimation [Internet]. IEEE SigPort; 2019. Available from : http://sigport.org/4073