A Study of the Human Perception of Synthetic Faces

Shen, Bingyu; RichardWebster, Brandon; O’Toole, Alice J.; Bowyer, Kevin W.; Scheirer, Walter J.

doi:10.48550/arxiv.2111.04230

Cited by 1 publication

(6 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We replicate an experiment carried out by Shen et al [46], in which subjects judge pairs of non-masked face images as fake or genuine, but we require subjects to annotate regions supporting their decisions. More specifically, participants were presented with a pair of face images (one being a synthetically-generated identity, and the other being an authentic facial image), and asked to decide which image is either the synthetic image or the real image in a two-alternative forced choice (2AFC) manner.…”

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

“…Human classification of face images as real/fake is only accurate when subjects annotate images. [46] found that humans did not accurately classify face images as real/fake (a). We repeated [46]'s study, adding a step in which the subjects annotated the image regions that support their classification.…”

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

“…[46] found that humans did not accurately classify face images as real/fake (a). We repeated [46]'s study, adding a step in which the subjects annotated the image regions that support their classification. In our study (b), human accuracy is substantially higher than in [46].…”

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

“…We repeated [46]'s study, adding a step in which the subjects annotated the image regions that support their classification. In our study (b), human accuracy is substantially higher than in [46]. lection, with green highlights corresponding to user annotated regions in an example pair, is presented in supplementary materials.…”

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

“…Given that the only difference between our protocol and that used in [46] was the annotation requirement, we can properly diagnose how annotating images impacts decision accuracy. Fig.…”

Section: Do Annotations Improve Human Accuracy?mentioning

confidence: 99%

See 4 more Smart Citations

CYBORG: Blending Human Saliency Into the Loss Improves Deep Learning

Boyd¹,

Tinsley²,

Bowyer³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Can deep learning models achieve greater generalization if their training is guided by reference to human perceptual abilities? And how can we implement this in a practical manner? This paper proposes a first-ever training strategy to ConveY Brain Oversight to Raise Generalization (CY-BORG). This new training approach incorporates humanannotated saliency maps into a CYBORG loss function that guides the model towards learning features from image regions that humans find salient when solving a given visual task. The Class Activation Mapping (CAM) mechanism is used to probe the model's current saliency in each training batch, juxtapose model saliency with human saliency, and penalize the model for large differences. Results on the task of synthetic face detection show that the CYBORG loss leads to significant improvement in performance on unseen samples consisting of face images generated from six Generative Adversarial Networks (GANs) across multiple classification network architectures. We also show that scaling to even seven times as much training data with standard loss cannot beat the accuracy of CYBORG loss. As a side effect, we observed that the addition of explicit region annotation to the task of synthetic face detection increased human classification performance. This work opens a new area of research on how to incorporate human visual saliency into loss functions. All data, code and pre-trained models used in this work are offered with this paper.

show abstract

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

Section: Acquisition Of Human Salient Regionsmentioning

confidence: 99%

“…Given that the only difference between our protocol and that used in [46] was the annotation requirement, we can properly diagnose how annotating images impacts decision accuracy. Fig.…”

Section: Do Annotations Improve Human Accuracy?mentioning

confidence: 99%

See 3 more Smart Citations