mlsanityData to 7 Oct 2026

Eyes Tell All: Irregular Pupil Shapes Reveal GAN-generated Faces

arXiv 2109.00162 ↗First posted 1 Sep 2021cs.CVcited by 48

Not on the map. Why?

Abstract (arXiv)

Generative adversary network (GAN) generated high-realistic human faces have been used as profile images for fake social media accounts and are visually challenging to discern from real ones. In this work, we show that GAN-generated faces can be exposed via irregular pupil shapes. This phenomenon is caused by the lack of physiological constraints in the GAN models. We demonstrate that such artifacts exist widely in high-quality GAN-generated faces and further describe an automatic method to extract the pupils from two eyes and analysis their shapes for exposing the GAN-generated faces. Qualitative and quantitative evaluations of our method suggest its simplicity and effectiveness in distinguishing GAN-generated faces.

Key facts. Cited by 48 papers in the map (data through 7 Oct 2026). Builds on 22 of its 40 references that are in the map. First posted on arXiv 1 Sep 2021.

Builds on

22 of 40 references.

References · 22

  1. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,415
  2. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
  3. 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,580
  4. 2017Progressive Growing of GANs for Improved Quality, Stability, and Variationcited by 2,967
  5. 2021Alias-Free Generative Adversarial Networkscited by 1,181
  6. 2019CNN-generated images are surprisingly easy to spot... for nowcited by 842
  7. 2018Exposing Deep Fakes Using Inconsistent Head Posescited by 294
  8. 2020Global Texture Enhancement for Fake Face Detection in the Wildcited by 220
  9. 2020Media Forensics and DeepFakes: an overviewcited by 208
  10. 2018Do GANs leave artificial fingerprints?cited by 188
  11. 2021Boundary IoU: Improving Object-Centric Image Segmentation Evaluationcited by 186
  12. 2018Detecting GAN-generated Imagery using Color Cuescited by 75
10 more
  1. 2020SofGAN: A Portrait Image Generator with Dynamic Stylingcited by 65
  2. 2020Exposing GAN-generated Faces Using Inconsistent Corneal Specular Highlightscited by 59
  3. 2019Incremental learning for the detection and classification of GAN-generated imagescited by 58
  4. 2020Detecting CNN-Generated Facial Images in Real-World Scenarioscited by 32
  5. 2019Exposing GAN-synthesized Faces Using Landmark Locationscited by 30
  6. 2020Learning by Minimizing the Sum of Ranked Rangecited by 27
  7. 2021TkML-AP: Adversarial Attacks to Top-k Multi-Label Learningcited by 26
  8. 2020Detection, Attribution and Localization of GAN Generated Imagescited by 24
  9. 2021Sum of Ranked Range Loss for Supervised Learningcited by 23
  10. 2018Identification of Deep Network Generated Images Using Disparities in Color Componentscited by 18

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 2; 2022: 8; 2023: 10; 2024: 11; 2025: 12; 2026: 5.

48 papers.

Sort:

By area:

  1. 2024AEROBLADE: Training-Free Detection of Latent Diffusion Images Using Autoencoder Reconstruction Errorcited by 123
  2. 2023Intriguing properties of synthetic images: from generative adversarial networks to diffusion modelscited by 121
  3. 2023Self-Supervised Video Forensics by Audio-Visual Anomaly Detectioncited by 84
  4. 2024Preserving Fairness Generalization in Deepfake Detectioncited by 70
  5. 2022SeeABLE: Soft Discrepancies and Bounded Contrastive Learning for Exposing Deepfakescited by 49
  6. 2023Improving Fairness in Deepfake Detectioncited by 40
  7. 2024Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Modelcited by 37
  8. 2023Quality-Agnostic Deepfake Detection with Intra-model Collaborative Learningcited by 32
  9. 2022GAN-generated Faces Detection: A Survey and New Perspectivescited by 29
  10. 2021Robust Attentive Deep Neural Network for Exposing GAN-generated Facescited by 28
  11. 2024Robust CLIP-Based Detector for Exposing Diffusion Model-Generated Imagescited by 20
  12. 2024Characteristics and prevalence of fake social media profiles with AI-generated facescited by 14
36 more
  1. 2025From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakescited by 14
  2. 2025Characterizing Photorealism and Artifacts in Diffusion Model-Generated Imagescited by 13
  3. 2023Synthesizing Black-box Anti-forensics DeepFakes with High Visual Qualitycited by 12
  4. 2022Open-Eye: An Open Platform to Study Human Performance on Identifying AI-Synthesized Facescited by 11
  5. 2022PseudoProp: Robust Pseudo-Label Generation for Semi-Supervised Object Detection in Autonomous Driving Systemscited by 10
  6. 2024AI-Generated Faces in the Real World: A Large-Scale Case Study of Twitter Profile Imagescited by 10
  7. 2024Masked Conditional Diffusion Model for Enhancing Deepfake Detectioncited by 9
  8. 2023Finding AI-Generated Faces in the Wildcited by 8
  9. 2024Semantic Contextualization of Face Forgery: A New Definition, Dataset, and Detection Methodcited by 8
  10. 2021Layered Controllable Video Generationcited by 7
  11. 2022An Overview of Recent Work in Media Forensics: Methods and Threatscited by 7
  12. 2023Harnessing the Power of Text-image Contrastive Models for Automatic Detection of Online Misinformationcited by 6
  13. 2023Evading Forensic Classifiers with Attribute-Conditioned Adversarial Facescited by 6
  14. 2022GOTCHA: Real-Time Video Deepfake Detection via Challenge-Responsecited by 5
  15. 2022A Dataless FaceSwap Detection Approach Using Synthetic Imagescited by 5
  16. 2025Robust AI-Generated Face Detection with Imbalanced Datacited by 5
  17. 2023X-Transfer: A Transfer Learning-Based Framework for GAN-Generated Fake Image Detectioncited by 4
  18. 2024Semantics-Oriented Multitask Learning for DeepFake Detection: A Joint Embedding Approachcited by 4
  19. 2025RAID: A Dataset for Testing the Adversarial Robustness of AI-Generated Image Detectorscited by 4
  20. 2025Rethinking Individual Fairness in Deepfake Detectioncited by 4
  21. 2025Bi-Level Optimization for Self-Supervised AI-Generated Face Detectioncited by 4
  22. 2024A Self-Learning Multimodal Approach for Fake News Detectioncited by 3
  23. 2025Self-Supervised Learning for Detecting AI-Generated Faces as Anomaliescited by 3
  24. 2025Generalizable AI-Generated Image Detection Based on Fractal Self-Similarity in the Spectrumcited by 2
  25. 2022Detection of Real-time DeepFakes in Video Conferencing with Active Probing and Corneal Reflectioncited by 1
  26. 2023Spritz-PS: Validation of Synthetic Face Images Using a Large Dataset of Printed Documentscited by 1
  27. 2024The Tug-of-War Between Deepfake Generation and Detectioncited by 1
  28. 2025DOOMGAN:High-Fidelity Dynamic Identity Obfuscation Ocular Generative Morphingcited by 1
  29. 2025PRADA: Probability-Ratio-Based Attribution and Detection of Autoregressive-Generated Imagescited by 1
  30. 2025Referee: Reference-aware Audiovisual Deepfake Detectioncited by 0
  31. 2025SynthGuard: An Open Platform for Detecting AI-Generated Multimedia with Multimodal LLMscited by 0
  32. 2026Color Matters: Demosaicing-Guided Color Correlation Training for Generalizable AI-Generated Image Detectioncited by 0
  33. 2026Moir'e Video Authentication: A Physical Signature Against AI Video Generationcited by 0
  34. 2026AIFIND: Artifact-Aware Interpreting Fine-Grained Alignment for Incremental Face Forgery Detectioncited by 0
  35. 2026When Eyes Betray AI: Social Gaze Consistency as a Semantic Cue for AI-Generated Image Detectioncited by 0
  36. 2026Training-Free Reconstruction-Based AI-Generated Image Detectors Are Inherently Vulnerable to Adversarial Examplescited by 0