A deep action-oriented video image classification system for text detection and recognition
For the video images with complex actions, achieving accurate text detection and recognition results is very challenging. This paper presents a hybrid model for classification of action-oriented video images which reduces the complexity of the problem to improve text detection and recognition perfor...
Saved in:
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Published: |
Springer
2021
|
Subjects: | |
Online Access: | http://eprints.um.edu.my/34804/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
id |
my.um.eprints.34804 |
---|---|
record_format |
eprints |
spelling |
my.um.eprints.348042022-09-07T06:32:27Z http://eprints.um.edu.my/34804/ A deep action-oriented video image classification system for text detection and recognition Chaudhuri, Abhra Shivakumara, Palaiahnakote Chowdhury, Pinaki Nath Pal, Umapada Lu, Tong Lopresti, Daniel Kumar, Govindaraj Hemantha QA75 Electronic computers. Computer science For the video images with complex actions, achieving accurate text detection and recognition results is very challenging. This paper presents a hybrid model for classification of action-oriented video images which reduces the complexity of the problem to improve text detection and recognition performance. Here, we consider the following five categories of genres, namely concert, cooking, craft, teleshopping and yoga. For classifying action-oriented video images, we explore ResNet50 for learning the general pixel-distribution level information and the VGG16 network is implemented for learning the features of Maximally Stable Extremal Regions and again another VGG16 is used for learning facial components obtained by a multitask cascaded convolutional network. The approach integrates the outputs of the three above-mentioned models using a fully connected neural network for classification of five action-oriented image classes. We demonstrated the efficacy of the proposed method by testing on our dataset and two other standard datasets, namely, Scene Text Dataset dataset which contains 10 classes of scene images with text information, and the Stanford 40 Actions dataset which contains 40 action classes without text information. Our method outperforms the related existing work and enhances the class-specific performance of text detection and recognition, significantly. Springer 2021-11 Article PeerReviewed Chaudhuri, Abhra and Shivakumara, Palaiahnakote and Chowdhury, Pinaki Nath and Pal, Umapada and Lu, Tong and Lopresti, Daniel and Kumar, Govindaraj Hemantha (2021) A deep action-oriented video image classification system for text detection and recognition. SN Applied Sciences, 3 (11). ISSN 2523-3971, DOI https://doi.org/10.1007/s42452-021-04821-z <https://doi.org/10.1007/s42452-021-04821-z>. 10.1007/s42452-021-04821-z |
institution |
Universiti Malaya |
building |
UM Library |
collection |
Institutional Repository |
continent |
Asia |
country |
Malaysia |
content_provider |
Universiti Malaya |
content_source |
UM Research Repository |
url_provider |
http://eprints.um.edu.my/ |
topic |
QA75 Electronic computers. Computer science |
spellingShingle |
QA75 Electronic computers. Computer science Chaudhuri, Abhra Shivakumara, Palaiahnakote Chowdhury, Pinaki Nath Pal, Umapada Lu, Tong Lopresti, Daniel Kumar, Govindaraj Hemantha A deep action-oriented video image classification system for text detection and recognition |
description |
For the video images with complex actions, achieving accurate text detection and recognition results is very challenging. This paper presents a hybrid model for classification of action-oriented video images which reduces the complexity of the problem to improve text detection and recognition performance. Here, we consider the following five categories of genres, namely concert, cooking, craft, teleshopping and yoga. For classifying action-oriented video images, we explore ResNet50 for learning the general pixel-distribution level information and the VGG16 network is implemented for learning the features of Maximally Stable Extremal Regions and again another VGG16 is used for learning facial components obtained by a multitask cascaded convolutional network. The approach integrates the outputs of the three above-mentioned models using a fully connected neural network for classification of five action-oriented image classes. We demonstrated the efficacy of the proposed method by testing on our dataset and two other standard datasets, namely, Scene Text Dataset dataset which contains 10 classes of scene images with text information, and the Stanford 40 Actions dataset which contains 40 action classes without text information. Our method outperforms the related existing work and enhances the class-specific performance of text detection and recognition, significantly. |
format |
Article |
author |
Chaudhuri, Abhra Shivakumara, Palaiahnakote Chowdhury, Pinaki Nath Pal, Umapada Lu, Tong Lopresti, Daniel Kumar, Govindaraj Hemantha |
author_facet |
Chaudhuri, Abhra Shivakumara, Palaiahnakote Chowdhury, Pinaki Nath Pal, Umapada Lu, Tong Lopresti, Daniel Kumar, Govindaraj Hemantha |
author_sort |
Chaudhuri, Abhra |
title |
A deep action-oriented video image classification system for text detection and recognition |
title_short |
A deep action-oriented video image classification system for text detection and recognition |
title_full |
A deep action-oriented video image classification system for text detection and recognition |
title_fullStr |
A deep action-oriented video image classification system for text detection and recognition |
title_full_unstemmed |
A deep action-oriented video image classification system for text detection and recognition |
title_sort |
deep action-oriented video image classification system for text detection and recognition |
publisher |
Springer |
publishDate |
2021 |
url |
http://eprints.um.edu.my/34804/ |
_version_ |
1744649197182779392 |
score |
13.154949 |