Natural and Effective Obfuscation by Head Inpainting

By Qianru Sun, Liqian Ma, Seong Joon Oh, Luc Van Gool, Bernt Schiele, Mario Fritz. 2017

226 citations recorded by the source corpus.

As more and more personal photos are shared online, being able to obfuscate identities in such photos is becoming a necessity for privacy protection. People have largely resorted to blacking out or blurring head regions, but they result in poor user experience while being surprisingly ineffective against state of the art person recognizers [17]. In this work, we propose a novel head inpainting obfuscation technique. Generating a realistic head inpainting in social media photos is challenging because subjects appear in diverse activities and head orientations. We thus split the task into two sub-tasks: (1) facial landmark generation from image context (e.g. body pose) for seamless hypothesis of sensible head pose, and (2) facial landmark conditioned head inpainting. We verify that our inpainting method generates realistic person images, while achieving superior obfuscation performance against automatic person recognizers.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

A Study of Face Obfuscation in ImageNet (2021, background): “Researchers have tried to protect sensitive image regions against attacks, e.g., by perturbing the image adversarially to reduce the performance of a recognizer (Oh et al., 2017; Ren et al., 2018; Sun et al., 2018; Wu et al., 2018; Xiao et al., 2020).”
A Survey on Differential Privacy for Unstructured Data Content (2022, methodology): “In [94], the GAN-based method was designed to inpaint head regions while protecting their naturalness.”
NAS-DIP: Learning Deep Image Prior with Neural Architecture Search (2020, background): “Among the design choices, 2D transposed convolution is one of the popular choices for various dense prediction tasks, including semantic segmentation [54], optical flow estimation [21], depth prediction [15], and image restoration [49, 73] and synthesis [59] problems.”
Fawkes: Protecting Privacy against Unauthorized Deep Learning Models (2020, background): “o edit facial images so that human-like characteristics are preserved but facial recognition model accuracy is signiﬁcantly reduced. Methods used includek-means facial averaging[35], facial inpainting[51], andGAN-basedface editing[27,52,64]. Since these dramatically alter the user’s face in her phot”
CIAGAN: Conditional Identity Anonymization Generative Adversarial Networks (2020, background): “This was previously explored in [29, 34, 14, 35, 7].”
LaFIn: Generative Landmark Guided Face Inpainting (2019, background): “Many works have successfully applied landmarks to the task of face generation, such as [36], [26] and [35].”
Live Face De-Identification in Video (2019, unknown): “In the GAN-based methods of [43, 44], face de-identification is employed for the related task of person obfuscation.”
Disentangled Person Image Generation (2017, unknown): “Facial landmarks can be transfered accordingly [31].”
When Machine Learning Meets Privacy (2020, background): “text, image, and video. Triastcyn et al. [156] used GAN to generating artificial data that retain statistical properties of the real data while reducing the risk of information disclosure. Sun et al. [149] proposed GAN-based head inpainting obfuscation technique to preserve the identity of users whe”
Celeb-DF: A Large-Scale Challenging Dataset for DeepFake Forensics (2019, methodology): “Instead, we follow the face in-painting work [45] and use the Mask-SSIM score [32] as a referenced quantitative metric of visual quality of synthesized DeepFake video frames.”
Knockoff Nets: Stealing Functionality of Black-Box Models (2018, background): “ries? 4.What makes for a good knockoff architecture? 2. Related Work Privacy, Security and Computer Vision. Privacy has been largely addressed within the computer vision community by proposing models [28,30,31,42,49,50] which recognize and control privacy-sensitive information in visual content. The”