GTSRB
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Image Classification | GTSRB | Accuracy99.8 | 291 | |
| Backdoor Defense | GTSRB (test) | ASR0 | 127 | |
| Image Classification | GTSRB | Natural Accuracy99.43 | 87 | |
| Image Classification | GTSRB | CA95.99 | 79 | |
| Image Classification | GTSRB (test) | Accuracy (Clean)97.85 | 59 | |
| Membership Inference Attack | GTSRB (test) | TPR@0.1%FPR43.52 | 42 | |
| Backdoor Detection | GTSRB | TPR100 | 39 | |
| Explanation Hardening Evaluation | GTSRB 2.5% utility loss, ϵ = 9, δ = 10⁻⁵ (test) | MLS (TPR@0.1% FPR) Pre0.39 | 30 | |
| Image Classification | GTSRB 0.5% BER (test) | CA (NF)96.5 | 30 | |
| Image Classification | GTSRB 1% BER (test) | CA (NF)96.4 | 30 | |
| Backdoor Defense | GTSRB 1% poison rate (test) | Clean Accuracy98 | 27 | |
| Image Classification | GTSRB Dp: ImageNet (test) | BA94.98 | 22 | |
| Image Classification | GTSRB Dp: CIFAR10 (test) | BA95.54 | 22 | |
| Backdoor Defense | GTSRB BadNets (test) | Attack Success Rate100 | 22 | |
| Backdoor Defense | GTSRB | PA91.73 | 21 | |
| Adversarial Robustness | GTSRB (test) | FGSM Accuracy74 | 18 | |
| Image Classification | GTSRB 32 x 32 43 classes (test) | Accuracy (CA)97.41 | 17 | |
| Backdoor Detection | GTSRB BppAttack (test) | AUC0.9967 | 15 | |
| Backdoor Detection | GTSRB WaNet attack (test) | AUC1 | 15 | |
| Backdoor Detection | GTSRB SIG attack (test) | AUC97.63 | 15 | |
| Watermark Detection | GTSRB | AccLoss12.29 | 14 | |
| Model Extraction Attack Robustness | GTSRB | Accuracy22.91 | 14 | |
| Backdoor Detection | GTSRB (test) | Top-1 ASR1 | 12 | |
| Backdoor Attack | GTSRB | Backdoor Accuracy80.35 | 12 | |
| Domain Adaptation Classification | Synthetic Signs → GTSRB (test) | Error Rate (%)1.82 | 10 |
Showing 25 of 64 rows