Spoofing 2D Face Detection: Machines See People Who Aren't There

TL;DR

This paper demonstrates that it is possible to create images that fool the Viola-Jones face detection algorithm, which humans do not recognize as faces, including printed and photographed versions, highlighting vulnerabilities in machine face detection.

Contribution

The study reveals that adversarial images can deceive Viola-Jones face detection without human recognition, including printed and photographed scenarios, exposing security risks.

Findings

Images can be generated that fool Viola-Jones without human face recognition.

Fooling images remain effective even after printing and photographing.

The work exposes vulnerabilities in 2D face detection algorithms.

Abstract

Machine learning is increasingly used to make sense of the physical world yet may suffer from adversarial manipulation. We examine the Viola-Jones 2D face detection algorithm to study whether images can be created that humans do not notice as faces yet the algorithm detects as faces. We show that it is possible to construct images that Viola-Jones recognizes as containing faces yet no human would consider a face. Moreover, we show that it is possible to construct images that fool facial detection even when they are printed and then photographed.