Enregistré dans:
Détails bibliographiques
Auteurs principaux: Sarkar, Ayush, Mai, Hanlin, Mahapatra, Amitabh, Lazebnik, Svetlana, Forsyth, D. A., Bhattad, Anand
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:https://arxiv.org/abs/2311.17138
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866929366651371520
author Sarkar, Ayush
Mai, Hanlin
Mahapatra, Amitabh
Lazebnik, Svetlana
Forsyth, D. A.
Bhattad, Anand
author_facet Sarkar, Ayush
Mai, Hanlin
Mahapatra, Amitabh
Lazebnik, Svetlana
Forsyth, D. A.
Bhattad, Anand
contents Generative models can produce impressively realistic images. This paper demonstrates that generated images have geometric features different from those of real images. We build a set of collections of generated images, prequalified to fool simple, signal-based classifiers into believing they are real. We then show that prequalified generated images can be identified reliably by classifiers that only look at geometric properties. We use three such classifiers. All three classifiers are denied access to image pixels, and look only at derived geometric features. The first classifier looks at the perspective field of the image, the second looks at lines detected in the image, and the third looks at relations between detected objects and shadows. Our procedure detects generated images more reliably than SOTA local signal based detectors, for images from a number of distinct generators. Saliency maps suggest that the classifiers can identify geometric problems reliably. We conclude that current generators cannot reliably reproduce geometric properties of real images.
format Preprint
id arxiv_https___arxiv_org_abs_2311_17138
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Shadows Don't Lie and Lines Can't Bend! Generative Models don't know Projective Geometry...for now
Sarkar, Ayush
Mai, Hanlin
Mahapatra, Amitabh
Lazebnik, Svetlana
Forsyth, D. A.
Bhattad, Anand
Computer Vision and Pattern Recognition
Artificial Intelligence
Graphics
Machine Learning
Generative models can produce impressively realistic images. This paper demonstrates that generated images have geometric features different from those of real images. We build a set of collections of generated images, prequalified to fool simple, signal-based classifiers into believing they are real. We then show that prequalified generated images can be identified reliably by classifiers that only look at geometric properties. We use three such classifiers. All three classifiers are denied access to image pixels, and look only at derived geometric features. The first classifier looks at the perspective field of the image, the second looks at lines detected in the image, and the third looks at relations between detected objects and shadows. Our procedure detects generated images more reliably than SOTA local signal based detectors, for images from a number of distinct generators. Saliency maps suggest that the classifiers can identify geometric problems reliably. We conclude that current generators cannot reliably reproduce geometric properties of real images.
title Shadows Don't Lie and Lines Can't Bend! Generative Models don't know Projective Geometry...for now
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Graphics
Machine Learning
url https://arxiv.org/abs/2311.17138