Face Detection in Distorted Images Based on Image Restoration and Meta-learning

Xinru Liu; Mingtao Pei; Wei Liang; Zhengang Nie

doi:10.1007/978-981-19-5096-4_13

Face Detection in Distorted Images Based on Image Restoration and Meta-learning

Xinru Liu, Mingtao Pei^*, Wei Liang, Zhengang Nie

^*Corresponding author for this work

Beijing Institute of Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Face detection in distorted images is a challenging task, and a natural idea is to conduct image restoration before face detection. Most of current image restoration techniques focus on improving the perceptual quality of the output image for single type of distortion, without taking the subsequent high-level detection task and unknown distortion into account. In this paper, we propose a restoration-based face detector in which the images are restored based on the detection loss instead of the perceptual quality loss, leading to better performance on subsequent detection task. Furthermore, we employ meta-learning to initialize the model with more appropriate parameters, thus our detector can adapt quickly to unseen distortions only using few examples with the corresponding distortion. Experiments on public datasets show that our proposed method could improve the performance of face detection in distorted images, and have a better generalization ability when applied to images with unseen distortions.

Original language	English
Title of host publication	Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers
Editors	Yongtian Wang, Huimin Ma, Yuxin Peng, Yue Liu, Ran He
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	171-180
Number of pages	10
ISBN (Print)	9789811950957
DOIs	https://doi.org/10.1007/978-981-19-5096-4_13
Publication status	Published - 2022
Event	17th Chinese Conference on Image and Graphics Technologies and Applications, IGTA 2022 - Virtual, Online Duration: 23 Apr 2022 → 24 Apr 2022

Publication series

Name	Communications in Computer and Information Science
Volume	1611 CCIS
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	17th Chinese Conference on Image and Graphics Technologies and Applications, IGTA 2022
City	Virtual, Online
Period	23/04/22 → 24/04/22

Keywords

Distorted images
Face detection
Image restoration
Meta-learning

Access to Document

10.1007/978-981-19-5096-4_13

Cite this

Liu, X., Pei, M., Liang, W., & Nie, Z. (2022). Face Detection in Distorted Images Based on Image Restoration and Meta-learning. In Y. Wang, H. Ma, Y. Peng, Y. Liu, & R. He (Eds.), Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers (pp. 171-180). (Communications in Computer and Information Science; Vol. 1611 CCIS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-19-5096-4_13

Liu, Xinru ; Pei, Mingtao ; Liang, Wei et al. / Face Detection in Distorted Images Based on Image Restoration and Meta-learning. Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers. editor / Yongtian Wang ; Huimin Ma ; Yuxin Peng ; Yue Liu ; Ran He. Springer Science and Business Media Deutschland GmbH, 2022. pp. 171-180 (Communications in Computer and Information Science).

@inproceedings{66e8292af9e24214847851547a8f1c87,

title = "Face Detection in Distorted Images Based on Image Restoration and Meta-learning",

abstract = "Face detection in distorted images is a challenging task, and a natural idea is to conduct image restoration before face detection. Most of current image restoration techniques focus on improving the perceptual quality of the output image for single type of distortion, without taking the subsequent high-level detection task and unknown distortion into account. In this paper, we propose a restoration-based face detector in which the images are restored based on the detection loss instead of the perceptual quality loss, leading to better performance on subsequent detection task. Furthermore, we employ meta-learning to initialize the model with more appropriate parameters, thus our detector can adapt quickly to unseen distortions only using few examples with the corresponding distortion. Experiments on public datasets show that our proposed method could improve the performance of face detection in distorted images, and have a better generalization ability when applied to images with unseen distortions.",

keywords = "Distorted images, Face detection, Image restoration, Meta-learning",

author = "Xinru Liu and Mingtao Pei and Wei Liang and Zhengang Nie",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; 17th Chinese Conference on Image and Graphics Technologies and Applications, IGTA 2022 ; Conference date: 23-04-2022 Through 24-04-2022",

year = "2022",

doi = "10.1007/978-981-19-5096-4_13",

language = "English",

isbn = "9789811950957",

series = "Communications in Computer and Information Science",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "171--180",

editor = "Yongtian Wang and Huimin Ma and Yuxin Peng and Yue Liu and Ran He",

booktitle = "Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers",

address = "Germany",

}

Liu, X, Pei, M , Liang, W & Nie, Z 2022, Face Detection in Distorted Images Based on Image Restoration and Meta-learning. in Y Wang, H Ma, Y Peng, Y Liu & R He (eds), Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers. Communications in Computer and Information Science, vol. 1611 CCIS, Springer Science and Business Media Deutschland GmbH, pp. 171-180, 17th Chinese Conference on Image and Graphics Technologies and Applications, IGTA 2022, Virtual, Online, 23/04/22. https://doi.org/10.1007/978-981-19-5096-4_13

Face Detection in Distorted Images Based on Image Restoration and Meta-learning. / Liu, Xinru; Pei, Mingtao ; Liang, Wei et al.
Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers. ed. / Yongtian Wang; Huimin Ma; Yuxin Peng; Yue Liu; Ran He. Springer Science and Business Media Deutschland GmbH, 2022. p. 171-180 (Communications in Computer and Information Science; Vol. 1611 CCIS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Face Detection in Distorted Images Based on Image Restoration and Meta-learning

AU - Liu, Xinru

AU - Pei, Mingtao

AU - Liang, Wei

AU - Nie, Zhengang

PY - 2022

Y1 - 2022

N2 - Face detection in distorted images is a challenging task, and a natural idea is to conduct image restoration before face detection. Most of current image restoration techniques focus on improving the perceptual quality of the output image for single type of distortion, without taking the subsequent high-level detection task and unknown distortion into account. In this paper, we propose a restoration-based face detector in which the images are restored based on the detection loss instead of the perceptual quality loss, leading to better performance on subsequent detection task. Furthermore, we employ meta-learning to initialize the model with more appropriate parameters, thus our detector can adapt quickly to unseen distortions only using few examples with the corresponding distortion. Experiments on public datasets show that our proposed method could improve the performance of face detection in distorted images, and have a better generalization ability when applied to images with unseen distortions.

AB - Face detection in distorted images is a challenging task, and a natural idea is to conduct image restoration before face detection. Most of current image restoration techniques focus on improving the perceptual quality of the output image for single type of distortion, without taking the subsequent high-level detection task and unknown distortion into account. In this paper, we propose a restoration-based face detector in which the images are restored based on the detection loss instead of the perceptual quality loss, leading to better performance on subsequent detection task. Furthermore, we employ meta-learning to initialize the model with more appropriate parameters, thus our detector can adapt quickly to unseen distortions only using few examples with the corresponding distortion. Experiments on public datasets show that our proposed method could improve the performance of face detection in distorted images, and have a better generalization ability when applied to images with unseen distortions.

KW - Distorted images

KW - Face detection

KW - Image restoration

KW - Meta-learning

UR - http://www.scopus.com/inward/record.url?scp=85135075152&partnerID=8YFLogxK

U2 - 10.1007/978-981-19-5096-4_13

DO - 10.1007/978-981-19-5096-4_13

M3 - Conference contribution

AN - SCOPUS:85135075152

SN - 9789811950957

T3 - Communications in Computer and Information Science

SP - 171

EP - 180

BT - Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers

A2 - Wang, Yongtian

A2 - Ma, Huimin

A2 - Peng, Yuxin

A2 - Liu, Yue

A2 - He, Ran

PB - Springer Science and Business Media Deutschland GmbH

T2 - 17th Chinese Conference on Image and Graphics Technologies and Applications, IGTA 2022

Y2 - 23 April 2022 through 24 April 2022

ER -

Liu X, Pei M , Liang W , Nie Z. Face Detection in Distorted Images Based on Image Restoration and Meta-learning. In Wang Y, Ma H, Peng Y, Liu Y, He R, editors, Image and Graphics Technologies and Applications - 17th Chinese Conference, IGTA 2022, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2022. p. 171-180. (Communications in Computer and Information Science). doi: 10.1007/978-981-19-5096-4_13