Generative-based Adversarial Attack Detection on ImageNet100 benchmark leaderboard · SOTA2 Research

Generative-based Adversarial Attack Detection on ImageNet100

0.9878CDAPerturbation Forgery

Evaluation Results

Method	Links
Perturbation Forgery2024.05		0.9878	0.9678	0.9364	0.9223	0.9223
EPSAD2024.05		0.9674	0.6997	0.7265	0.47	0.1507
SPAD2024.05		0.9385	0.9064	0.8984	0.8862	0.8879
DIRE2024.05		0.8976	0.9097	0.9129	0.9097	0.9143
Universal-Detector2024.05		0.7945	0.817	0.8312	0.9202	0.5531
CNN-Detection2024.05		0.7051	0.6743	0.6917	0.3963	0.526
LGrad2024.05		0.6144	0.6077	0.6068	0.5586	0.5835