Sign Language Recognition on LSA 64 (test)
100Top-1 AccuracySPOTER
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SPOTERInput Modality=Skeleton-based, Source=Original authors2025.03 | 100 | — | — | — | |
| SiformerInput Modality=Skeleton-based2025.03 | 99.84 | — | — | — | |
| SPOTERInput Modality=Skeleton-based, Source=Reproduction2025.03 | 99.52 | — | — | — | |
| MEMPInput Modality=RGB-based2025.03 | 99.06 | — | — | — | |
| facebook/timesformer-base-finetuned-k400Backbone=TimeSformer-base, Finetuned=K4002025.06 | 99.06 | 99.2 | 99.1 | 99.1 | |
| I3DInput Modality=RGB-based2025.03 | 98.91 | — | — | — | |
| Hierarchical Windowed Graph Attention Network2025.06 | 98.59 | — | — | — | |
| LSTM + LDSInput Modality=RGB-based2025.03 | 98.09 | — | — | — | |
| MCG-NJU/videomae-base-finetuned-kineticsBackbone=VideoMAE-base, Finetuning=Kinetics2025.06 | 97.65 | 98.3 | 97.6 | 97.5 | |
| google/vivit-b-16x2-kinetics400Backbone=ViViT-B-16x2, Pre-training=Kinetics4002025.06 | 97.18 | 97.7 | 97.2 | 97.1 | |
| MCG-NJU/videomae-baseBackbone=VideoMAE-base2025.06 | 96.25 | 96.8 | 96.3 | 96.1 | |
| DeepSign CNNInput Modality=RGB-based2025.03 | 96 | — | — | — | |
| HMM-GMM2025.06 | 95.95 | — | — | — | |
| 3D Graph Convolutional Neural Network2025.06 | 94.84 | — | — | — | |
| LSTM + DSCInput Modality=Skeleton-based2025.03 | 92.15 | — | — | — |