Protect, show, attend and tell: Empowering image captioning models with ownership protection
By and large, existing Intellectual Property (IP) protection on deep neural networks typically i) focus on image classification task only, and ii) follow a standard digital watermarking framework that was conventionally used to protect the ownership of multimedia and video content. This paper demons...
Saved in:
Main Authors: | , , , , |
---|---|
Format: | Article |
Published: |
Elsevier Sci Ltd
2022
|
Subjects: | |
Online Access: | http://eprints.um.edu.my/33782/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
id |
my.um.eprints.33782 |
---|---|
record_format |
eprints |
spelling |
my.um.eprints.337822022-04-26T08:10:05Z http://eprints.um.edu.my/33782/ Protect, show, attend and tell: Empowering image captioning models with ownership protection Lim, Jian Han Chan, Chee Seng Ng, Kam Woh Fan, Lixin Yang, Qiang QA75 Electronic computers. Computer science TA Engineering (General). Civil engineering (General) By and large, existing Intellectual Property (IP) protection on deep neural networks typically i) focus on image classification task only, and ii) follow a standard digital watermarking framework that was conventionally used to protect the ownership of multimedia and video content. This paper demonstrates that the current digital watermarking framework is insufficient to protect image captioning tasks that are often regarded as one of the frontiers AI problems. As a remedy, this paper studies and proposes two different embedding schemes in the hidden memory state of a recurrent neural network to protect the image captioning model. From empirical points, we prove that a forged key will yield an unusable image captioning model, defeating the purpose of infringement. To the best of our knowledge, this work is the first to propose ownership protection on image captioning task. Also, extensive experiments show that the proposed method does not compromise the original image captioning performance on all common captioning metrics on Flickr30k and MS-COCO datasets, and at the same time it is able to withstand both removal and ambiguity attacks. Code is available at https://github.com/jianhanlim/ipr-imagecaptioning (c) 2021 Elsevier Ltd. All rights reserved. Elsevier Sci Ltd 2022-02 Article PeerReviewed Lim, Jian Han and Chan, Chee Seng and Ng, Kam Woh and Fan, Lixin and Yang, Qiang (2022) Protect, show, attend and tell: Empowering image captioning models with ownership protection. Pattern Recognition, 122. ISSN 0031-3203, DOI https://doi.org/10.1016/j.patcog.2021.108285 <https://doi.org/10.1016/j.patcog.2021.108285>. 10.1016/j.patcog.2021.108285 |
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 TA Engineering (General). Civil engineering (General) |
spellingShingle |
QA75 Electronic computers. Computer science TA Engineering (General). Civil engineering (General) Lim, Jian Han Chan, Chee Seng Ng, Kam Woh Fan, Lixin Yang, Qiang Protect, show, attend and tell: Empowering image captioning models with ownership protection |
description |
By and large, existing Intellectual Property (IP) protection on deep neural networks typically i) focus on image classification task only, and ii) follow a standard digital watermarking framework that was conventionally used to protect the ownership of multimedia and video content. This paper demonstrates that the current digital watermarking framework is insufficient to protect image captioning tasks that are often regarded as one of the frontiers AI problems. As a remedy, this paper studies and proposes two different embedding schemes in the hidden memory state of a recurrent neural network to protect the image captioning model. From empirical points, we prove that a forged key will yield an unusable image captioning model, defeating the purpose of infringement. To the best of our knowledge, this work is the first to propose ownership protection on image captioning task. Also, extensive experiments show that the proposed method does not compromise the original image captioning performance on all common captioning metrics on Flickr30k and MS-COCO datasets, and at the same time it is able to withstand both removal and ambiguity attacks. Code is available at https://github.com/jianhanlim/ipr-imagecaptioning (c) 2021 Elsevier Ltd. All rights reserved. |
format |
Article |
author |
Lim, Jian Han Chan, Chee Seng Ng, Kam Woh Fan, Lixin Yang, Qiang |
author_facet |
Lim, Jian Han Chan, Chee Seng Ng, Kam Woh Fan, Lixin Yang, Qiang |
author_sort |
Lim, Jian Han |
title |
Protect, show, attend and tell: Empowering image captioning models with ownership protection |
title_short |
Protect, show, attend and tell: Empowering image captioning models with ownership protection |
title_full |
Protect, show, attend and tell: Empowering image captioning models with ownership protection |
title_fullStr |
Protect, show, attend and tell: Empowering image captioning models with ownership protection |
title_full_unstemmed |
Protect, show, attend and tell: Empowering image captioning models with ownership protection |
title_sort |
protect, show, attend and tell: empowering image captioning models with ownership protection |
publisher |
Elsevier Sci Ltd |
publishDate |
2022 |
url |
http://eprints.um.edu.my/33782/ |
_version_ |
1735409590810443776 |
score |
13.18916 |