Semantic Segmentation on Embryo image dataset (test)
94.51Accuracy (%)SplitFed-CL
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SplitFed-CLLabel Setting=Noisy Labels, Mode=Full2026.05 | 94.51 | 0.0725 | — | 82.65 | 78.62 | 87.8 | 89.23 | 89.95 | |
| SplitFed-CLLabel Setting=All Accurate Labels2026.05 | 94.48 | 0.0753 | — | 81.47 | 78.16 | 88.11 | 88.98 | 89.6 | |
| SplitFed-CLLabel Setting=Noisy Labels, Ablation=No Consistency Loss2026.05 | 94.34 | 0.0775 | — | 82.5 | 76.51 | 87.05 | 87.99 | 89.33 | |
| FedNCL-V2Label Setting=Noisy Labels2026.05 | 94.2 | 0.0776 | — | 82.81 | 78.28 | 87.02 | 88.14 | 89.07 | |
| FedAVGLabel Setting=All Accurate Labels2026.05 | 93.68 | 0.096 | — | 76.84 | 73.39 | 85.76 | 86.05 | 86.61 | |
| ARFLLabel Setting=Noisy Labels2026.05 | 93.65 | 0.083 | — | 82.12 | 77.13 | 85.73 | 87.32 | 88.39 | |
| Smart SplitFedsigma_noise=02026.01 | 93.6 | 0.06 | 94 | 80 | 75 | 85 | 88 | — | |
| Smart SplitFedsigma_noise=2 * 10^-42026.01 | 93.52 | 0.07 | 95 | 78 | 73 | 83 | 88 | — | |
| Naive SplitFedsigma_noise=02026.01 | 93.39 | 0.08 | 93 | 78 | 74 | 85 | 88 | — | |
| SplitFed-CLLabel Setting=Noisy Labels, Ablation=No Label Correction2026.05 | 93.26 | 0.0806 | — | 81.06 | 75.33 | 86.12 | 86.55 | 88.25 | |
| Smart SplitFedsigma_noise=6 * 10^-42026.01 | 93.18 | 0.07 | 94 | 77 | 72 | 83 | 87 | — | |
| Smart SplitFedsigma_noise=5 * 10^-12026.01 | 93.12 | 0.06 | 94 | 78 | 73 | 82 | 87 | — | |
| QA-SplitFedLabel Setting=Noisy Labels2026.05 | 93.12 | 0.0945 | — | 80.77 | 73.18 | 84.6 | 86.32 | 86.95 | |
| Smart SplitFedsigma_noise=10^-22026.01 | 93.05 | 0.08 | 94 | 79 | 74 | 85 | 88 | — | |
| Smart SplitFedsigma_noise=10^-32026.01 | 92.97 | 0.12 | 93 | 78 | 74 | 83 | 87 | — | |
| FedMixLabel Setting=Noisy Labels2026.05 | 92.89 | 0.1033 | — | 77.8 | 70.47 | 84.23 | 85.14 | 86.08 | |
| Smart SplitFedsigma_noise=10^-12026.01 | 92.66 | 0.08 | 93 | 78 | 74 | 85 | 86 | — | |
| SplitFedAVGsigma_noise=02026.01 | 92.5 | 0.1 | 92 | 75 | 74 | 83 | 87 | — | |
| FedAVGLabel Setting=Noisy Labels2026.05 | 92.34 | 0.1074 | — | 78.83 | 68.46 | 83.72 | 84.05 | 85.57 | |
| DHLCLabel Setting=Noisy Labels2026.05 | 92.32 | 0.1349 | — | 72.05 | 60.37 | 81.85 | 84.16 | 82.25 | |
| CELCLabel Setting=Noisy Labels2026.05 | 91.78 | 0.1387 | — | 72.96 | 58.46 | 81.56 | 83.15 | 82.09 | |
| SplitFedAVGsigma_noise=2 * 10^-42026.01 | 89.44 | 0.13 | 88 | 71 | 71 | 77 | 83 | — | |
| Naive SplitFedsigma_noise=2 * 10^-42026.01 | 89 | 0.17 | 90 | 71 | 70 | 73 | 79 | — | |
| SplitFedAVGsigma_noise=6 * 10^-42026.01 | 87.1 | 0.16 | 86 | 64 | 65 | 74 | 79 | — | |
| Naive SplitFedsigma_noise=6 * 10^-42026.01 | 82.02 | 0.21 | 74 | 51 | 65 | 75 | 79 | — | |
| SplitFedAVGsigma_noise=10^-32026.01 | 76.49 | 0.23 | 59 | 50 | 67 | 73 | 78 | — | |
| Naive SplitFedsigma_noise=10^-32026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — | |
| Naive SplitFedsigma_noise=10^-22026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — | |
| SplitFedAVGsigma_noise=10^-22026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — | |
| Naive SplitFedsigma_noise=10^-12026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — | |
| SplitFedAVGsigma_noise=10^-12026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — | |
| Naive SplitFedsigma_noise=5 * 10^-12026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — | |
| SplitFedAVGsigma_noise=5 * 10^-12026.01 | 40.27 | — | 40 | 0 | 0 | 0 | 0 | — |