Protecting Image Processing Networks via Model Watermarking

Jie Zhang; Dongdong Chen; Jing Liao; Weiming Zhang; Nenghai Yu

doi:10.1007/978-981-19-7554-7_6

Protecting Image Processing Networks via Model Watermarking

Jie Zhang^*
, Dongdong Chen
, Jing Liao
, Weiming Zhang
, Nenghai Yu

^*Corresponding author for this work

Department of Computer Science

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 12 - Chapter in an edited book (Author) › peer-review

Abstract

Deep learning has achieved tremendous success in low-level computer vision tasks such as image processing tasks. To protect the intellectual property (IP) of such valuable image processing networks, the model vendor can sell the service in the manner of the application program interface (API). However, even if the attacker can only query the API, he is still able to conduct model extraction attacks, which can steal the functionality of the target networks. In this chapter, we propose a new model watermarking framework for image processing networks. Under the framework, two strategies are further developed, namely, the model-agnostic strategy and the model-specific strategy. The proposed watermarking method performs well in terms of fidelity, capacity, and robustness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023

Original language	English
Title of host publication	Digital Watermarking for Machine Learning Model
Subtitle of host publication	Techniques, Protocols and Applications
Editors	Lixin Fan, Chee Seng Chan, Qiang Yang
Place of Publication	Singapore
Publisher	Springer
Chapter	6
Pages	93–116
ISBN (Electronic)	978-981-19-7554-7
ISBN (Print)	978-981-19-7553-0
DOIs	https://doi.org/10.1007/978-981-19-7554-7_6
Publication status	Published - 2023

Bibliographical note

Research Unit(s) information for this publication is provided by the author(s) concerned.

Funding

This research was partly supported by the Natural Science Foundation of China under Grant U20B2047, 62072421, 62002334, 62102386, and 62121002, Exploration Fund Project of University of Science and Technology of China under Grant YD3480002001.

Access Output

10.1007/978-981-19-7554-7_6

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@inbook{3163b70bc7174b9a89241e107ab9e0b7,

title = "Protecting Image Processing Networks via Model Watermarking",

abstract = "Deep learning has achieved tremendous success in low-level computer vision tasks such as image processing tasks. To protect the intellectual property (IP) of such valuable image processing networks, the model vendor can sell the service in the manner of the application program interface (API). However, even if the attacker can only query the API, he is still able to conduct model extraction attacks, which can steal the functionality of the target networks. In this chapter, we propose a new model watermarking framework for image processing networks. Under the framework, two strategies are further developed, namely, the model-agnostic strategy and the model-specific strategy. The proposed watermarking method performs well in terms of fidelity, capacity, and robustness. {\textcopyright} The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023",

author = "Jie Zhang and Dongdong Chen and Jing Liao and Weiming Zhang and Nenghai Yu",

note = "Research Unit(s) information for this publication is provided by the author(s) concerned.",

year = "2023",

doi = "10.1007/978-981-19-7554-7\_6",

language = "English",

isbn = "978-981-19-7553-0",

pages = "93–116",

editor = "Lixin Fan and Chan, \{Chee Seng\} and Qiang Yang",

booktitle = "Digital Watermarking for Machine Learning Model",

publisher = "Springer ",

}

Protecting Image Processing Networks via Model Watermarking. / Zhang, Jie; Chen, Dongdong; Liao, Jing et al.
Digital Watermarking for Machine Learning Model: Techniques, Protocols and Applications. ed. / Lixin Fan; Chee Seng Chan; Qiang Yang. Singapore: Springer , 2023. p. 93–116.

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 12 - Chapter in an edited book (Author) › peer-review

TY - CHAP

T1 - Protecting Image Processing Networks via Model Watermarking

AU - Zhang, Jie

AU - Chen, Dongdong

AU - Liao, Jing

AU - Zhang, Weiming

AU - Yu, Nenghai

N1 - Research Unit(s) information for this publication is provided by the author(s) concerned.

PY - 2023

Y1 - 2023

N2 - Deep learning has achieved tremendous success in low-level computer vision tasks such as image processing tasks. To protect the intellectual property (IP) of such valuable image processing networks, the model vendor can sell the service in the manner of the application program interface (API). However, even if the attacker can only query the API, he is still able to conduct model extraction attacks, which can steal the functionality of the target networks. In this chapter, we propose a new model watermarking framework for image processing networks. Under the framework, two strategies are further developed, namely, the model-agnostic strategy and the model-specific strategy. The proposed watermarking method performs well in terms of fidelity, capacity, and robustness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023

AB - Deep learning has achieved tremendous success in low-level computer vision tasks such as image processing tasks. To protect the intellectual property (IP) of such valuable image processing networks, the model vendor can sell the service in the manner of the application program interface (API). However, even if the attacker can only query the API, he is still able to conduct model extraction attacks, which can steal the functionality of the target networks. In this chapter, we propose a new model watermarking framework for image processing networks. Under the framework, two strategies are further developed, namely, the model-agnostic strategy and the model-specific strategy. The proposed watermarking method performs well in terms of fidelity, capacity, and robustness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023

UR - https://www.scopus.com/pages/publications/85172155132

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85172155132&origin=recordpage

U2 - 10.1007/978-981-19-7554-7_6

DO - 10.1007/978-981-19-7554-7_6

M3 - RGC 12 - Chapter in an edited book (Author)

SN - 978-981-19-7553-0

SP - 93

EP - 116

BT - Digital Watermarking for Machine Learning Model

A2 - Fan, Lixin

A2 - Chan, Chee Seng

A2 - Yang, Qiang

PB - Springer

CY - Singapore

ER -

Protecting Image Processing Networks via Model Watermarking

Abstract

Bibliographical note

Funding

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this