Detecting Photoshopped Faces by Scripting Photoshop

By Sheng-Yu Wang, Oliver Wang, Andrew Owens, Richard Zhang, Alexei A. Efros. 2019

152 citations recorded by the source corpus.

Most malicious photo manipulations are created using standard image editing tools, such as Adobe Photoshop. We present a method for detecting one very popular Photoshop manipulation -- image warping applied to human faces -- using a model trained entirely using fake images that were automatically generated by scripting Photoshop itself. We show that our model outperforms humans at the task of recognizing manipulated images, can predict the specific location of edits, and in some cases can be used to "undo" a manipulation to reconstruct the original, unedited image. We demonstrate that the system can be successfully applied to artist-created image manipulations.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

CNN-Generated Images Are Surprisingly Easy to Spot… for Now (2019, methodology): “We have experimented with running our detector on the face-aware liquify dataset from [38], and found that our method performs at chance on this data.”
Towards Universal Fake Image Detectors that Generalize Across Generative Models (2023, methodology): “Some could be natural images that are doctored using tools like Adobe Photoshop [1, 48], while others could have been generated through a machine learning algorithm.”
Multi-attentional Deepfake Detection (2021, unknown): “Many works [46, 23, 4, 53, 34, 22, 33, 26, 45, 43] have been proposed.”
On the Detection of Digital Face Manipulation (2019, background): “FaceSwap can replace the identity of 3 D models while preserving the original expressions.”
DeepFakes and Beyond: A Survey of Face Manipulation and Fake Detection (2020, methodology): “Wang et al. carried out in [80] an interesting research using publicly available commercial software from Adobe Photo-shop (Face-Aware Liquify tool [100]) in order to synthesise new faces, and also a professional artist in order to manipulate 50 real photographs.”
The Stable Signature: Rooting Watermarks in Latent Diffusion Models (2023, background): “One way is to detect inconsistencies in the generated images, via lights, perspective or physical objects [21, 22, 45, 52, 88].”
Steering Self-Supervised Feature Learning Beyond Local Pixel Statistics (2020, background): “[49] predict subtle face image manipulations based on local warping.”
GANprintR: Improved Fakes and Evaluation of the State of the Art in Face Manipulation Detection (2019, background): “carried out in [17] a very interesting research using”
DIRE for Diffusion-Generated Image Detection (2023, methodology): “Following previous generated-image detection methods [47, 46, 51], we report accuracy (ACC) and average precision (AP) in our experiments to evaluate the detectors.”
FakeSpotter: A Simple Baseline for Spotting AI-Synthesized Fake Faces (2019, methodology): “Adobe proposes a method to detect manipulated images with Photo-shop which also manipulates images with computer graphics techniques [19].”
Media Forensics and DeepFakes: An Overview (2020, methodology): “Finally, in [134] a CNN based solution is devised to detect artifacts introduced by a specific Photoshop tool, Face-Aware Liquify, which performs image warping of human faces.”