Handwritten Character Classification on EMNIST-Digits (test)
99.82Accuracyµ2Net
Evaluation results
| Method | Links | |
|---|---|---|
| µ2NetIteration=1st, Knowledge Reuse=false2022.05 | 99.82 | |
| µ2NetIteration=2nd, Knowledge Reuse=false2022.05 | 99.82 | |
| WaveMixAugmentation=TrivialAugment2022.05 | 99.82 | |
| Previous SOTA [75]2022.05 | 99.82 | |
| WaveMix-128/7Number of parameters (Million)=2.42022.03 | 99.77 | |
| Jeevan & Sethi (2022)2022.05 | 99.77 | |
| WaveMix-256/7Number of parameters (Million)=9.62022.03 | 99.7 | |
| ResNet-34Number of parameters (Million)=21.32022.03 | 99.68 | |
| ResNet-18Number of parameters (Million)=11.22022.03 | 99.67 | |
| ResNet-50Number of parameters (Million)=23.62022.03 | 99.62 | |
| Dufourq et al.Training set size=Full2019.04 | 99.3 | |
| Cohen et al.Training set size=Full2019.04 | 95.9 |