Large-scale privacy protection in Google Street View

By Andrea Frome, German K. M. Cheung, Ahmad Abdulkader, Marco Zennaro, Bo Wu, Alessandro Bissacco, Hartwig Adam, Hartmut Neven, Luc Vincent. 2009

284 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

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

In Signal Processing

Citations in the corpus

Privacy in Deep Learning: A Survey (2020, background): “This poses serious privacy concerns, as neural networks are used in different aspects of our lives [5, 3, 6, 7, 8, 9, 10].”
Defeating Image Obfuscation with Deep Learning (2016, background): “Google Street View blurs license plates and faces [10].”
A Study of Face Obfuscation in ImageNet (2021, methodology): “Frome et al. (2009) and Uittenbogaard et al. (2019) use blurring and inpainting to obfuscate faces and license plates in Google Street View. nuScenes (Caesar et al., 2020) is an autonomous driving dataset where faces and license plates are detected and then blurred.”
Large-Scale FPGA-based Convolutional Networks (2011, background): “The approach was very successful, and led to several implementations, ranging from optical character recognition (OCR) to object detection, scene segmentation, and robot navigation: • check reading (handwriting recognition) at AT&T (LeCun et al., 1998b) and Microsoft (Simard et al., 2003; Chellapill”
A survey of the recent architectures of deep convolutional neural networks (2019, background): “nt in image sensors for face detection in video conferencing, and regulation of street crimes, etc. [76]–[78]. Likewise, CNN based systems were industrialized in markets for customers’ tracking [14], [79], [80]. Moreover, CNN’s potential in other applications such as medical image segmentation, anom”
A Survey on Differential Privacy for Unstructured Data Content (2022, methodology): “For instance, the box blurring algorithm was introduced in the privacy-protecting Google street view to protect human faces and vehicle license plates in [38].”
Conditional (2019, unknown): “As the earliest and most straightforward approaches, blurring smooths out the facial features across the imageHarmon & Julesz (1973); Frome et al. (2009), while pixelation involves enlarging the pixels to a size that obscures detailZhou & Pun (2020).”
Pose-Sensitive Embedding by Nonlinear NCA Regression (2010, background): “In our domain, hands might only occupy a few pixels, and the only body-part that can reliably be detected is the human face ([26, 13]).”
Highly Efficient Forward and Backward Propagation of Convolutional Neural Networks for Pixelwise Classification (2014, methodology): “In recent years, CNN has also been applied to object detection [15], [6], [18], [17], [13], image segmentation [20], scene labeling [5], [14], and tracking [4], and signiﬁcantly improved the accuracies in these applications.”
Evaluation of Pooling Operations in Convolutional Architectures for Object Recognition (2010, background): “They are deployed in commercial systems to read checks [10] and to obfuscate faces and license plates in Google StreetView [5] .”
Reading Digits in Natural Images with Unsupervised Feature Learning (2011, methodology): “We use a sliding-window classiﬁer based on a set of elementary features computed from image intensities and gradients (see [12] for details).”