HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images

Thumbnail image of item number 1 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Thumbnail image of item number 2 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Thumbnail image of item number 3 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Thumbnail image of item number 4 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Showing 1-4 of 13 pages in this article.

PDF Version Also Available for Download.

Description

This is the Accepted Manuscript version of an article that proposes HAR-Depth with sequential and shape learning along with the novel concept of depth history image (DHI) to address the challenges of Human action recognition (HAR). Results suggest that the proposed work of this paper performs better in terms of overall accuracy, kappa parameter and precision compared to the other state-of-the-art algorithms present in the earlier reported literature.

Physical Description

13 p.

Creation Information

Sahoo, Suraj Prakash; Ari, Samit; Mahapatra, Kamalakanta & Mohanty, Saraju P. August 24, 2020.

Context

This article is part of the collection entitled: UNT Scholarly Works and was provided by the UNT College of Engineering to the UNT Digital Library, a digital repository hosted by the UNT Libraries. It has been viewed 30 times. More information about this article can be viewed below.

Who

People and organizations associated with either the creation of this article or its content.

Authors

Publisher

Provided By

UNT College of Engineering

The UNT College of Engineering strives to educate and train engineers and technologists who have the vision to recognize and solve the problems of society. The college comprises six degree-granting departments of instruction and research.

About | Browse this Partner

Contact Us

Corrections Questions

What

Descriptive information to help identify this article. Follow the links below to find similar items on the Digital Library.

Degree Information

Description

This is the Accepted Manuscript version of an article that proposes HAR-Depth with sequential and shape learning along with the novel concept of depth history image (DHI) to address the challenges of Human action recognition (HAR). Results suggest that the proposed work of this paper performs better in terms of overall accuracy, kappa parameter and precision compared to the other state-of-the-art algorithms present in the earlier reported literature.

Physical Description

13 p.

Notes

Abstract: Human action recognition (HAR) is a challenging task due to the presence of the pose and temporal variations in the action videos. To address these challenges, HAR-Depth is proposed in this paper with sequential and shape learning along with the novel concept of depth history image (DHI). A deep bidirectional long short term memory (DBiLSTM) is constructed for sequential learning to model the temporal relationship existing between the action frames. Action information in each frame is extracted using pre-trained convolutional neural network (CNN). The depth information of each action frame is estimated and projected onto the X-Y plane to form the DHI. During shape learning, the shape information through DHI is used to train a deep pre-trained CNN network. By leveraging the trained knowledge of the pre-trained network, overfitting issue is handled. The finetuned network is used to recognize actions from query DHI images. Data augmentation is adopted to avoid overfitting of the network by virtually increasing the training set. The proposed work is evaluated on publicly available datasets like KTH, UCF sports, JHMDB, UCF101, and HMDB51 and achieves the performance accuracy of 97.67%, 95.00%, 73.13%, 92.97%, and 69.74% respectively. The results on these datasets suggest that the proposed work of this paper performs better in terms of overall accuracy, kappa parameter and precision compared to the other state-of-the-art algorithms present in the earlier reported literature.

This is the Accepted Manuscript version of the article. The final version is available at: https://doi.org/10.1109/TETCI.2020.3014367

Source

  • IEEE Transactions on Emerging Topics in Computational Intelligence, 5(5), Institute of Electrical and Electronics Engineers, August 24, 2020, pp. 1-13

Language

Item Type

Identifier

Unique identifying numbers for this article in the Digital Library or other systems.

Publication Information

  • Publication Title: IEEE Transactions on Emerging Topics in Computational Intelligence
  • Volume: 5
  • Issue: 5
  • Page Start: 813
  • Page End: 825
  • Peer Reviewed: Yes

Collections

This article is part of the following collection of related materials.

UNT Scholarly Works

Materials from the UNT community's research, creative, and scholarly activities and UNT's Open Access Repository. Access to some items in this collection may be restricted.

About | Browse this Collection

What responsibilities do I have when using this article?

Digital Files

When

Dates and time periods associated with this article.

Creation Date

  • August 24, 2020

Added to The UNT Digital Library

  • Feb. 21, 2022, 1:06 p.m.

Description Last Updated

  • Dec. 4, 2023, 11:25 a.m.

Usage Statistics

When was this article last used?

Yesterday: 0
Past 30 days: 1
Total Uses: 30
More Statistics

Interact With This Article

Here are some suggestions for what to do next.

Start Reading

Thumbnail image of item number 1 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Thumbnail image of item number 2 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Thumbnail image of item number 3 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.
Thumbnail image of item number 4 in: 'HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images'.

PDF Version Also Available for Download.

Citations, Rights, Re-Use

International Image Interoperability Framework

IIF Logo

We support the IIIF Presentation API

Links for Robots

Helpful links in machine-readable formats.

Archival Resource Key (ARK)

International Image Interoperability Framework (IIIF)

Metadata Formats

Images

URLs

Stats

Sahoo, Suraj Prakash; Ari, Samit; Mahapatra, Kamalakanta & Mohanty, Saraju P. HAR-Depth: A Novel Framework for Human Action Recognition Using Sequential Learning and Depth Estimated History Images, article, August 24, 2020; (https://digital.library.unt.edu/ark:/67531/metadc1913264/: accessed May 1, 2024), University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu; crediting UNT College of Engineering.

Back to Top of Screen