Large image datasets: A pyrrhic win for computer vision?

In this paper we investigate problematic practices and consequences of large scale vision datasets (LSVDs). We examine broad issues such as the question of consent and justice as well as specific concerns such as the inclusion of verifiably pornographic images in datasets. Taking the ImageNet-ILSVRC-2012 dataset as an example, we perform a cross-sectional model-based quantitative census covering factors such as age, gender, NSFW content scoring, class- wise accuracy, human-cardinality-analysis, and the semanticity of the image class information in order to statistically investigate the extent and subtleties of ethical transgressions. We then use the census to help hand-curate a look-up-table of images in the ImageNet-ILSVRC-2012 dataset that fall into the categories of verifiably pornographic: shot in a non-consensual setting (up-skirt), beach voyeuristic, and exposed private parts. We survey the landscape of harm and threats both the society at large and individuals face due to uncritical and ill-considered dataset curation practices. We then propose possible courses of correction and critique their pros and cons. We have duly open-sourced all of the code and the census meta-datasets generated in this endeavor for the computer vision community to build on. By unveiling the severity of the threats, our hope is to motivate the constitution of mandatory Institutional Review Boards (IRB) for large scale dataset curation.

[1]  Rediet Abebe,et al.  Fairness, Equality, and Power in Algorithmic Decision-Making , 2021, FAccT.

[2]  Natalia Kovalyova,et al.  Data feminism , 2020, Information, Communication & Society.

[3]  Xiaohua Zhai,et al.  Are we done with ImageNet? , 2020, ArXiv.

[4]  Aleksander Madry,et al.  From ImageNet to Image Classification: Contextualizing Progress on Benchmarks , 2020, ICML.

[5]  Tom B. Brown,et al.  Measuring the Algorithmic Efficiency of Neural Networks , 2020, ArXiv.

[6]  C. Rudin,et al.  PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models , 2020, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[7]  S. Merz.  Race after technology. Abolitionist tools for the new Jim Code , 2020, Ethnic and Racial Studies.

[8]  Alexander Wong,et al.  Investigating the Impact of Inclusion in Face Recognition Training Data on Individual Face Identification , 2020, AIES.

[9]  Fei-Fei Li,et al.  Towards fairer datasets: filtering and balancing the distribution of the people subtree in the ImageNet hierarchy , 2019, FAT*.

[10]  Abeba Birhane,et al.  Algorithmic Injustices: Towards a Relational Ethics , 2019, ArXiv.

[11]  Jed R. Brubaker,et al.  How Computers See Gender , 2019, Proc. ACM Hum. Comput. Interact..

[12]  Yi Chern Tan,et al.  Assessing Social and Intersectional Biases in Contextualized Word Representations , 2019, NeurIPS.

[13]  Time to discuss consent in digital-data studies , 2019, Nature.

[14]  Luc Rocher,et al.  Estimating the success of re-identifications in incomplete datasets using generative models , 2019, Nature Communications.

[15]  Luciano Floridi,et al.  Translating Principles into Practices of Digital Ethics: Five Risks of Being Unethical , 2019, Philosophy & Technology.

[16]  Vinay Uday Prabhu,et al.  Fonts-2-Handwriting: A Seed-Augment-Train framework for universal digit classification , 2019, ArXiv.

[17]  Mary L. Gray,et al.  Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass , 2019 .

[18]  Alexander Wong,et al.  Auditing ImageNet: Towards a Model-driven Framework for Annotating Demographic Attributes of Large-Scale Image Datasets , 2019, ArXiv.

[19]  Stefanos Zafeiriou,et al.  RetinaFace: Single-stage Dense Face Localisation in the Wild , 2019, ArXiv.

[20]  Eric P. Xing,et al.  Learning Robust Global Representations by Penalizing Local Predictive Power , 2019, NeurIPS.

[21]  Claus Aranha,et al.  Data Augmentation Using GANs , 2019, ArXiv.

[22]  Kinga Polynczuk-Alenius,et al.  Algorithms of oppression: how search engines reinforce racism , 2019, Information, Communication & Society.

[23]  A. Gillespíe,et al.  Tackling Voyeurism: Is the Voyeurism (Offences) Act 2019 a Wasted Opportunity? , 2019, The Modern Law Review.

[24]  Aaron Hertzmann,et al.  Aesthetics of Neural Network Art , 2019, ArXiv.

[25]  Yoav Goldberg,et al.  Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Them , 2019, NAACL.

[26]  Judy Hoffman,et al.  Predictive Inequity in Object Detection , 2019, ArXiv.

[27]  Benjamin Recht,et al.  Do ImageNet Classifiers Generalize to ImageNet? , 2019, ICML.

[28]  Baoyuan Wu,et al.  Tencent ML-Images: A Large-Scale Multi-Label Image Database for Visual Representation Learning , 2019, IEEE Access.

[29]  A. Hanbury,et al.  Measuring Societal Biases in Text Corpora via First-Order Co-occurrence , 2018 .

[30]  Alexei A. Efros,et al.  Dataset Distillation , 2018, ArXiv.

[31]  Hannah Lebovits.  Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor , 2018, Public Integrity.

[32]  Inioluwa Deborah Raji,et al.  Model Cards for Model Reporting , 2018, FAT.

[33]  Jeff Donahue,et al.  Large Scale GAN Training for High Fidelity Natural Image Synthesis , 2018, ICLR.

[34]  Liyue Fan,et al.  Image Pixelization with Differential Privacy , 2018, DBSec.

[35]  Kate Saenko,et al.  VisDA: A Synthetic-to-Real Benchmark for Visual Domain Adaptation , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[36]  Bolei Zhou,et al.  Places: A 10 Million Image Database for Scene Recognition , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[37]  Luc Van Gool,et al.  Deep Expectation of Real and Apparent Age from a Single Image Without Facial Landmarks , 2016, International Journal of Computer Vision.

[38]  Timnit Gebru,et al.  Datasheets for datasets , 2018, Commun. ACM.

[39]  Vijayan K. Asari,et al.  The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches , 2018, ArXiv.

[40]  Leland McInnes,et al.  UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction , 2018, ArXiv.

[41]  S. Zafeiriou,et al.  ArcFace: Additive Angular Margin Loss for Deep Face Recognition , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[42]  Timnit Gebru,et al.  Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification , 2018, FAT.

[43]  Moustapha Cissé,et al.  ConvNets and ImageNet Beyond Accuracy: Understanding Mistakes and Uncovering Biases , 2017, ECCV.

[44]  D. Sculley,et al.  No Classification without Representation: Assessing Geodiversity Issues in Open Data Sets for the Developing World , 2017, 1711.08536.

[45]  Jaakko Lehtinen,et al.  Progressive Growing of GANs for Improved Quality, Stability, and Variation , 2017, ICLR.

[46]  Tony Doyle,et al.  Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy , 2017, Inf. Soc..

[47]  Erika Rackley,et al.  More than 'Revenge Porn' : image-based sexual abuse and the reform of Irish law. , 2017 .

[48]  Vitaly Shmatikov,et al.  Machine Learning Models that Remember Too Much , 2017, CCS.

[49]  C. McGlynn,et al.  Image-based sexual abuse. , 2017 .

[50]  Chen Sun,et al.  Revisiting Unreasonable Effectiveness of Data in Deep Learning Era , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[51]  Stephanie Baran,et al.  Visual patriarchy: PETA advertising and the commodification of sexualized bodies , 2017 .

[52]  C. McGlynn,et al.  Beyond ‘Revenge Porn’: The Continuum of Image-Based Sexual Abuse , 2017 .

[53]  François Chollet,et al.  Xception: Deep Learning with Depthwise Separable Convolutions , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[54]  J. Callahan,et al.  Gender and Musical Instrument Stereotypes in Middle School Children , 2016 .

[55]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[56]  Sean A. Munson,et al.  Unequal Representation and Gender Stereotypes in Image Search Results for Occupations , 2015, CHI.

[57]  Geoffrey E. Hinton,et al.  Distilling the Knowledge in a Neural Network , 2015, ArXiv.

[58]  Jeffrey Pennington,et al.  GloVe: Global Vectors for Word Representation , 2014, EMNLP.

[59]  Michael S. Bernstein,et al.  ImageNet Large Scale Visual Recognition Challenge , 2014, International Journal of Computer Vision.

[60]  Alexander L. Brown,et al.  Why Do People Volunteer? An Experimental Analysis of Preferences for Time Donations , 2013, Manag. Sci..

[61]  Pedro F. Miret,et al.  Wikipedia , 2008, Monatsschrift für Deutsches Recht.

[62]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[63]  Jessica M. Coates,et al.  The School Girl, the Billboard, and Virgin: The Virgin Mobile Case and the Use of Creative Commons Licensed Photographs by Commercial Entities , 2011 .

[64]  Susan Corbett,et al.  Creative Commons Licences, the Copyright Regime and the Online Community: Is There a Fatal Disconnect? , 2011 .

[65]  Herkko Hietanen,et al.  Creative Commons Olympics How Big Media is Learning to License from Amateur Authors , 2011 .

[66]  A. Powell.  Configuring Consent: Emerging Technologies, Unauthorized Sexual Images and Sexual Assault , 2010 .

[67]  S. Naidoo,et al.  Informed consent for photography in dental practice : communication , 2009 .

[68]  Fei-Fei Li,et al.  ImageNet: A large-scale hierarchical image database , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[69]  Margaret A. Boden,et al.  What is generative art? , 2009, Digit. Creativity.

[70]  Susan Corbett,et al.  Creative Commons Licences: A Symptom or a Cause? , 2009 .

[71]  Antonio Torralba,et al.  Ieee Transactions on Pattern Analysis and Machine Intelligence 1 80 Million Tiny Images: a Large Dataset for Non-parametric Object and Scene Recognition , 2022 .

[72]  Vitaly Shmatikov,et al.  Robust De-anonymization of Large Sparse Datasets , 2008, 2008 IEEE Symposium on Security and Privacy (sp 2008).

[73]  Geoffrey C. Bowker,et al.  Enacting silence: Residual categories as a challenge for ethics, information systems, and communication , 2007, Ethics and Information Technology.

[74]  Lucy Suchman.  Human-Machine Reconfigurations: Plans and Situated Actions , 2006 .

[75]  Michael Ramirez,et al.  “My Dog's Just Like Me”: Dog Ownership as a Gender Display , 2006 .

[76]  Robert Eaglestone,et al.  One and the Same? Ethics, Aesthetics, and Truth , 2004 .

[77]  J. Tanner,et al.  Informed Consent the Global Picture , 2002, British journal of perioperative nursing : the journal of the National Association of Theatre Nurses.

[78]  J. Overhage,et al.  Sorting Things Out: Classification and Its Consequences , 2001, Annals of Internal Medicine.

[79]  Paul Weindling,et al.  The Origins of Informed Consent: The International Scientific Commission on Medical War Crimes, and the Nuremberg Code , 2001, Bulletin of the history of medicine.

[80]  Judith M. Tanur,et al.  Gender and Musical Instruments: Winds of Change? Jason Zervoudakes , 1994 .

[81]  E. Hirschman.  Consumers and Their Animal Companions , 1994 .

[82]  Judith K. Delzell,et al.  Gender Association of Musical Instruments and Preferences of Fourth-Grade Students for Selected Instruments , 1992 .

[83]  Seymour A. Papert,et al.  The Summer Vision Project , 1966 .

[84]  R. Marshall.  Automating Inequality. How High-Tech Tools Profile, Police and Punish the Poor , 2018 .

[85]  Asher Flynn,et al.  Not Just 'Revenge Pornography': Australians' Experiences of Image-Based Abuse: A Summary Report , 2017 .

[86]  Maiko Spiess.  Sorting Things Out - Classification and Its Consequences , 2010 .

[87]  Manamai Ozaki,et al.  Shashinjinsei: Nobuyoshi Araki's Photo Journey Art and Not or Pornography , 2008 .

[88]  Julie L. Fishman.  Is Diamond Smuggling Forever? The Kimberley Process Certification Scheme: The First Step Down the Long Road to Solving the Blood Diamond Trade Problem , 2005 .

[89]  Lawrence Lessig,et al.  The Creative Commons , 2004 .

[90]  D. E. Rogers,et al.  Where have we been? Where are we going? , 1986, Daedalus.