Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences
This paper proposes an action recognition framework for depth map sequences using the 3D Space-Time Auto-Correlation of Gradients (STACOG) algorithm. First, each depth map sequence is split into two sets of sub-sequences of two different frame lengths individually. Second, a number of Depth Motion M...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-05-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/21/11/3642 |
id |
doaj-8e24baa35daf44329130c0f3f0703fb2 |
---|---|
record_format |
Article |
spelling |
doaj-8e24baa35daf44329130c0f3f0703fb22021-06-01T00:55:54ZengMDPI AGSensors1424-82202021-05-01213642364210.3390/s21113642Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps SequencesMohammad Farhad Bulbul0Sadiya Tabussum1Hazrat Ali2Wenli Zheng3Mi Young Lee4Amin Ullah5Department of Mathematics, Jashore University of Science and Technology, Jashore 7408, BangladeshDepartment of Mathematics, Jashore University of Science and Technology, Jashore 7408, BangladeshDepartment of Electrical and Computer Engineering, Abbottabad Campus, COMSATS University Islamabad, Abbottabad 22060, PakistanSchool of Science, Xi’an Shiyou University, Xi’an 710065, ChinaIntelligent Media Laboratory, Department of Software, Sejong University, Seoul 143-747, KoreaIntelligent Media Laboratory, Department of Software, Sejong University, Seoul 143-747, KoreaThis paper proposes an action recognition framework for depth map sequences using the 3D Space-Time Auto-Correlation of Gradients (STACOG) algorithm. First, each depth map sequence is split into two sets of sub-sequences of two different frame lengths individually. Second, a number of Depth Motion Maps (DMMs) sequences from every set are generated and are fed into STACOG to find an auto-correlation feature vector. For two distinct sets of sub-sequences, two auto-correlation feature vectors are obtained and applied gradually to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>L</mi><mn>2</mn></msub></semantics></math></inline-formula>-regularized Collaborative Representation Classifier (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>L</mi><mn>2</mn></msub></semantics></math></inline-formula>-CRC) for computing a pair of sets of residual values. Next, the Logarithmic Opinion Pool (LOGP) rule is used to combine the two different outcomes of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>L</mi><mn>2</mn></msub></semantics></math></inline-formula>-CRC and to allocate an action label of the depth map sequence. Finally, our proposed framework is evaluated on three benchmark datasets named MSR-action 3D dataset, DHA dataset, and UTD-MHAD dataset. We compare the experimental results of our proposed framework with state-of-the-art approaches to prove the effectiveness of the proposed framework. The computational efficiency of the framework is also analyzed for all the datasets to check whether it is suitable for real-time operation or not.https://www.mdpi.com/1424-8220/21/11/36423D action recognitiondepth motion maps3D auto-correlation featuresdecision fusionRegularized Collaborative Representation Classifier (CRC) |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Mohammad Farhad Bulbul Sadiya Tabussum Hazrat Ali Wenli Zheng Mi Young Lee Amin Ullah |
spellingShingle |
Mohammad Farhad Bulbul Sadiya Tabussum Hazrat Ali Wenli Zheng Mi Young Lee Amin Ullah Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences Sensors 3D action recognition depth motion maps 3D auto-correlation features decision fusion Regularized Collaborative Representation Classifier (CRC) |
author_facet |
Mohammad Farhad Bulbul Sadiya Tabussum Hazrat Ali Wenli Zheng Mi Young Lee Amin Ullah |
author_sort |
Mohammad Farhad Bulbul |
title |
Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences |
title_short |
Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences |
title_full |
Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences |
title_fullStr |
Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences |
title_full_unstemmed |
Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences |
title_sort |
exploring 3d human action recognition using stacog on multi-view depth motion maps sequences |
publisher |
MDPI AG |
series |
Sensors |
issn |
1424-8220 |
publishDate |
2021-05-01 |
description |
This paper proposes an action recognition framework for depth map sequences using the 3D Space-Time Auto-Correlation of Gradients (STACOG) algorithm. First, each depth map sequence is split into two sets of sub-sequences of two different frame lengths individually. Second, a number of Depth Motion Maps (DMMs) sequences from every set are generated and are fed into STACOG to find an auto-correlation feature vector. For two distinct sets of sub-sequences, two auto-correlation feature vectors are obtained and applied gradually to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>L</mi><mn>2</mn></msub></semantics></math></inline-formula>-regularized Collaborative Representation Classifier (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>L</mi><mn>2</mn></msub></semantics></math></inline-formula>-CRC) for computing a pair of sets of residual values. Next, the Logarithmic Opinion Pool (LOGP) rule is used to combine the two different outcomes of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>L</mi><mn>2</mn></msub></semantics></math></inline-formula>-CRC and to allocate an action label of the depth map sequence. Finally, our proposed framework is evaluated on three benchmark datasets named MSR-action 3D dataset, DHA dataset, and UTD-MHAD dataset. We compare the experimental results of our proposed framework with state-of-the-art approaches to prove the effectiveness of the proposed framework. The computational efficiency of the framework is also analyzed for all the datasets to check whether it is suitable for real-time operation or not. |
topic |
3D action recognition depth motion maps 3D auto-correlation features decision fusion Regularized Collaborative Representation Classifier (CRC) |
url |
https://www.mdpi.com/1424-8220/21/11/3642 |
work_keys_str_mv |
AT mohammadfarhadbulbul exploring3dhumanactionrecognitionusingstacogonmultiviewdepthmotionmapssequences AT sadiyatabussum exploring3dhumanactionrecognitionusingstacogonmultiviewdepthmotionmapssequences AT hazratali exploring3dhumanactionrecognitionusingstacogonmultiviewdepthmotionmapssequences AT wenlizheng exploring3dhumanactionrecognitionusingstacogonmultiviewdepthmotionmapssequences AT miyounglee exploring3dhumanactionrecognitionusingstacogonmultiviewdepthmotionmapssequences AT aminullah exploring3dhumanactionrecognitionusingstacogonmultiviewdepthmotionmapssequences |
_version_ |
1721413534200365056 |