Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

Illustration of human-AI perception discrepancy in alpha channel attacks.

More »

Fig 1 Expand

Fig 2.

Comparative evaluation of adversarial attack methods across four key criteria.

More »

Fig 2 Expand

Table 1.

Model architectures, training parameters, and hardware settings.

More »

Table 1 Expand

Fig 3.

Visual deception of alpha channel attack in industrial inspection.

More »

Fig 3 Expand

Fig 4.

Output deviation of Fast-GAN under alpha channel attack.

More »

Fig 4 Expand

Fig 5.

Training flowchart of YOLO detection model under poisoned data.

More »

Fig 5 Expand

Fig 6.

Detection failure of YOLO model induced by alpha channel attack.

More »

Fig 6 Expand

Fig 7.

Output deviation of DeepSeek-VL on alpha attacked image.

More »

Fig 7 Expand

Fig 8.

Misinterpretation of alpha attacked image by ChatGPT-4.

More »

Fig 8 Expand

Fig 9.

Misunderstanding of alpha attacked image by Kimi multimodal model.

More »

Fig 9 Expand

Table 2.

Detection performance (mAP@50/ mAP@50–95) of each defect category under different attack methods.

More »

Table 2 Expand

Fig 10.

Impact of alpha-channel attack on GAN training (measured by FID score).

More »

Fig 10 Expand

Table 3.

Language consistency evaluation under alpha-channel attacks.

More »

Table 3 Expand

Table 4.

Multimodal alignment degradation measured by CLIP score.

More »

Table 4 Expand

Table 5.

Semantic drift in text descriptions caused by alpha-channel attacks.

More »

Table 5 Expand

Table 6.

Comparative impact of alpha channel, FGSM, and boundary attacks across diverse models and tasks.

More »

Table 6 Expand

Fig 11.

Flow chart of alpha channel attack image detection.

More »

Fig 11 Expand

Fig 12.

Detection pipeline and quantitative metrics for alpha channel attack.

More »

Fig 12 Expand

Fig 13.

ROC curve for alpha channel attack detection.

More »

Fig 13 Expand

Table 7.

Performance metrics of the alpha-channel-based detection model.

More »

Table 7 Expand

Table 8.

Performance comparison of main adversarial defense methods across detection quality, efficiency, and robustness dimensions.

More »

Table 8 Expand

Fig 14.

Adversarially perturbed weld defect dataset.

More »

Fig 14 Expand

Fig 15.

Grayscale histogram of weld data under attack.

More »

Fig 15 Expand

Fig 16.

MSE heatmap of weld data under attack.

More »

Fig 16 Expand

Fig 17.

Adversarially perturbed PCB defect dataset.

More »

Fig 17 Expand

Fig 18.

Grayscale histogram of PCB data under attack.

More »

Fig 18 Expand

Fig 19.

MSE heatmap of PCB data under attack.

More »

Fig 19 Expand

Fig 20.

Adversarially perturbed steel surface defect dataset.

More »

Fig 20 Expand

Fig 21.

Grayscale histogram of steel surface data under attack.

More »

Fig 21 Expand

Fig 22.

MSE heatmap of steel surface data under attack.

More »

Fig 22 Expand