Papers of the day   All papers

Fooling automated surveillance cameras: adversarial patches to attack person detection

Comments

Pascal Bornet: New approach to fool #MachineLearning, avoiding face recognition https://arxiv.org/abs/1904.08653 @alvinfoo @FrRonconi @MikeQuindazzi @Paula_Piccard @ronald_vanloon @jerome_joffre @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen @samiranghosh @YuHelenYu @MHiesboeck @andy_lucerne https://t.co/k2VDugdtMe

5 replies, 152 likes


Olivier Laborde at #WebSummit2019: New approach to fool #MachineLearning, avoiding face recognition http://arxiv.org/abs/1904.08653 TY @pascal_bornet @3itcom @ipfconline1 @MikeQuindazzi @Paula_Piccard @AkwyZ @kalydeoo @jblefevre60 @Ym78200 @mvollmer1 @HeinzvHoenen @Fabriziobustama @YuHelenYu @enricomolinari @andy_lucerne https://t.co/xuj8WqGzsc

1 replies, 46 likes


Darshan H Sheth ✨ @Iamdarshan: New approach to fool #MachineLearning, avoiding face recognition https://arxiv.org/abs/1904.08653 @pascal_bornet @darshan_h_sheth @alvinfoo @FrRonconi @MikeQuindazzi @Paula_Piccard @ronald_vanloon @jerome_joffre @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen https://t.co/RVYjhf6G5G

2 replies, 36 likes


Darshan H Sheth ✨ @Iamdarshan: New approach to fool #MachineLearning, avoiding face recognition https://arxiv.org/abs/1904.08653 @pascal_bornet @darshan_h_sheth @alvinfoo @FrRonconi @MikeQuindazzi @Paula_Piccard @ronald_vanloon @jerome_joffre @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen https://t.co/RVYjhf6G5G

4 replies, 32 likes


Chris Heilmann: "How to hide from the AI surveillance state with a color printout" https://www.technologyreview.com/f/613409/how-to-hide-from-the-ai-surveillance-state-with-a-color-printout/ In-depth paper here: https://arxiv.org/abs/1904.08653

0 replies, 26 likes


Clifford Wolf: 10 years from now the equivalence of the tide pot challenge will be daring people to cross the street dressed like that

0 replies, 23 likes


hardmaru 😷: The above figure is from this paper that proposed a way of generating adversarial patches against YOLOv2: https://twitter.com/hardmaru/status/1120132384026808326

1 replies, 21 likes


hardmaru: Current deep learning gradient-based solutions probably won’t help. e.g. 👇🏼 https://twitter.com/hardmaru/status/1120132384026808326?s=21

0 replies, 21 likes


will knight: A few months ago, I made a t-shirt featuring an image that fools certain AI person detectors (see https://arxiv.org/pdf/1904.08653.pdf). Unlike the one mentioned in @jackclarkSF's latest newsletter, it is not robust to fabric deformations, so the wearer needs washboard abs. https://t.co/KCFJ6ie9II

0 replies, 20 likes


Chet Faliszek: This is really cool... I love the strange arms race we are in.

1 replies, 20 likes


Manoj Rawat: Yes! There are always numerous solutions to any technology. @alvinfoo @FrRonconi @MikeQuindazzi @Paula_Piccard @ronald_vanloon @jerome_joffre @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen @samiranghosh @YuHelenYu @MHiesboeck @andy_lucerne

1 replies, 13 likes


Davor Jordacevic: A few months ago, the University of KU Leuven in Belgium has worked on a new approach to fool surveillance and face recognition cameras by wearing an adversarial patch. From @pascal_bornet on LinkedIn

0 replies, 12 likes


Diogo Mónica: New business model for academic researchers: 1) Find optimal adversarial patterns 2) Sell a subscription-based t-shirt service.

1 replies, 8 likes


dvorkin ⭕: This is too cool: A 40cm-square patch that renders you invisible to person-detecting AIs. #AI #MachineVision https://boingboing.net/2019/04/25/tee-of-invisibility.html Original paper is here: https://arxiv.org/pdf/1904.08653.pdf

1 replies, 8 likes


Montreal.AI: "Fooling automated surveillance cameras: adversarial patches to attack person detection" Thys et al.: https://arxiv.org/abs/1904.08653 #ArtificialIntelligence #ComputerVision #PatternRecognition https://t.co/rbzlLdHAHo

0 replies, 7 likes


Naomi Saphra: Robustness guaranteed: this image classifier is vulnerable only to VERY UNFASHIONABLE adversaries.

0 replies, 6 likes


Frank Denis: RT @hardmaru: We're living in a cyberpunk future: “Fooling automated surveillance cameras: adversarial patches to attack person detection” https://arxiv.org/abs/1904.08653 https://t.co/OvHFRd4AoK

0 replies, 6 likes


Martijn Rasser: @CNAStech A1. Interesting paper on fooling automated surveillance cameras: 1/2 #CNASxTech #tech https://arxiv.org/abs/1904.08653

1 replies, 5 likes


american psycho: Ok this is the clothing brand I am doing We will deploy East german raindrop camo combined with high fashion to defeat the machines! Stay tuned @heywildrich

1 replies, 4 likes


Rimah Harb: #Adversarial patches help disappear from #video surveillance and face #recognition https://arxiv.org/abs/1904.08653 @alvinfoo @kashthefuturist @FrRonconi @MikeQuindazzi @Paula_Piccard @ronald_vanloon @jerome_joffre @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen @robvank @YuHelenYu

0 replies, 4 likes


Praveen Dalal: Fooling automated surveillance cameras: adversarial patches to attack person detection (Pdf) https://arxiv.org/pdf/1904.08653.pdf. When State is @_Orwellian, privacy and civil liberties have to be protected using innovative #legaltech and other ideas. @_Perry4Law @_TeleLaws @DigitalIndiaLaw https://t.co/oB4BLXB2Pc

0 replies, 2 likes


James W Breaux: https://arxiv.org/pdf/1904.08653.pdf A patch to make you invisible ... to the machines.

1 replies, 2 likes


Harvard Data Science: Weekend Paper Fooling automated surveillance cameras: adversarial patches to attack person detection "From our results we can see that our system is able significantly lower the accuracy of a person detector." https://arxiv.org/pdf/1904.08653.pdf

0 replies, 2 likes


thor magnusson: The t-shirt design of the 21st century! #surveillance #AI #MachineLearning https://arxiv.org/pdf/1904.08653.pdf https://t.co/Laoy2IBwKw

0 replies, 2 likes


Osumi, Yusuke: I wonder adversarial attacks on machine learning models are *true* cyber defense in these days. They succeeded in fooling cameras with a simple panel. Interesting. "Fooling automated surveillance cameras: adversarial patches to attack person detection" https://arxiv.org/abs/1904.08653 https://t.co/ICltDKhm1E

0 replies, 1 likes


Craig Iskowitz: I may started walking around with this photo around my waist anytime I go out.

0 replies, 1 likes


WSI-eMarketing: This is so interesting!!! How to fool #MachineLearning to avoid #facialrecognition

0 replies, 1 likes


Ektropos: How to cheat #AI powered public cameras evade surveillance and avoid detection and face recognition, without wearing a mask. #MachineLearning #Anarchy

0 replies, 1 likes


Alex Petric: Feeling paranoid? You should be! Cameras are tracking us. Get a "do-not-track" T-Shirt: https://stealth.cool powered by the work of: Simen Thys, Wiebe Van Ranst https://arxiv.org/pdf/1904.08653.pdf #DoNotTrack #SpreadTheWord

1 replies, 1 likes


Praveen Dalal: @DigitalIndiaLaw Automated surveillance cameras can also be fooled using many methods https://twitter.com/AadhaarScrap/status/1191213840425529344. The #legaltech and civil liberties advocate must encourage the legitimate use of these methods world over. @_Perry4Law @FacialMatch @MyDataIsMine @DigitalIndiaLaw

0 replies, 1 likes


Enrico Glerean: CC talks about a recent work on how to fool computer vision algorithms by wearing pictures https://arxiv.org/abs/1904.08653 #hbd19 https://t.co/s7MpHBVRvD

1 replies, 1 likes


AInara Bilbao 🤖: adversarial patches to attack person detection #MachineLearning via http://twinybots.ch https://arxiv.org/abs/1904.08653

0 replies, 1 likes


Covered Dish People: #1yrago Angered by the No-More-AOCs rule, 31 colleges' Young Democrats boycott the DCCC https://docs.google.com/document/d/1fsJdaCpqlTfwexwS2hZ0YxfLAfUz5xvM7WQWdqrROKQ/edit?fbclid=IwAR2rL7-6fMamnPlDmJgYt6MzsN0EIVMbIemv2V8KD-yDdSwDsl5Hk0hZL0o #1yrago A 40cm-square patch that renders you invisible to person-detecting AIs https://arxiv.org/abs/1904.08653 16/ https://t.co/gRsBAztwrk

1 replies, 0 likes


Content

Found on Nov 03 2019 at https://arxiv.org/pdf/1904.08653.pdf

PDF content of a computer science paper: Fooling automated surveillance cameras: adversarial patches to attack person detection