Fingerspelling Recognition on ChicagoFSWild (test)
72.9Letter AccuracyHandReader_RGB+KP
Evaluation Results
| Method | Links | |
|---|---|---|
| HandReader_RGB+KPModality=RGB + KP, Number of convolution modules=22025.05 | 72.9 | |
| HandReader_RGBModality=RGB2025.05 | 72 | |
| HandReader_KPModality=KP2025.05 | 69.3 | |
| Pose-based transformerModality=KP2025.05 | 66.3 | |
| Iter train + SupConModality=RGB2025.05 | 64.9 | |
| SiamseNet+Iteraive modelModality=RGB2025.05 | 49.6 | |
| 3d hand pose + RGBModality=RGB + KP2025.05 | 47.9 | |
| Iterative AttentionModality=RGB2025.05 | 45.1 | |
| Hand Det + CNN + RNNModality=RGB2025.05 | 41.9 |