Sign Language Translation on PHOENIX-2014T (dev)
35.1BLEU-4 ScoreADAT
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ADATstrategy=HATL2026.02 | 35.1 | 54.5 | 70.7 | 51 | 38.1 | — | — | |
| SignXCategory=Sign2Text2025.04 | 30.08 | 55.32 | 55.24 | 43.1 | 35.45 | — | — | |
| CorrNet+Category=Sign2Text2025.04 | 29.13 | 54.54 | 54.56 | 42.31 | 34.48 | — | — | |
| Two Stream-SLT2026.02 | 28.7 | 54.1 | 54.3 | 42 | 34.2 | — | — | |
| TwoStream-SLTLLM=true2023.08 | 28.66 | 54.08 | 54.32 | 41.99 | 34.15 | — | — | |
| TS-SLTSetting=Gloss-based2024.04 | 28.66 | 54.08 | 54.32 | 41.99 | 34.15 | — | — | |
| TwoStream-SLTInput Modality=RGB2024.05 | 28.66 | 54.08 | 54.32 | 41.99 | 34.15 | — | — | |
| TwoStream-SLTCategory=Sign2Text2025.04 | 28.66 | 54.08 | 54.32 | 41.99 | 34.15 | — | — | |
| TS-SLTModality (Pose)=true, Modality (RGB)=true, Approach=Gloss-based2026.06 | 28.66 | 54.08 | 54.32 | — | — | — | — | |
| ADATstrategy=HATL2026.02 | 28.6 | 54.6 | 60.5 | 50.8 | 38 | — | — | |
| IP-SLTLLM=false2023.08 | 28.22 | 54.43 | 54.1 | 41.56 | 33.66 | — | — | |
| IP-SLTInput Modality=RGB2024.05 | 28.22 | 54.43 | 54.1 | 41.56 | 33.66 | — | — | |
| IP-SLTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-based2026.06 | 28.22 | 54.43 | 54.1 | — | — | — | — | |
| TwoStream-SLTInput Modality=Keypoint2024.05 | 28.1 | 53.32 | 53.66 | 41.31 | 33.55 | — | — | |
| SLTUNET2026.02 | 27.9 | 52.2 | — | — | — | — | — | |
| SLTUNETSetting=Gloss-based2024.04 | 27.87 | 52.23 | — | — | — | — | — | |
| SLTUNETCategory=Sign2Text2025.04 | 27.87 | 52.23 | — | — | — | — | — | |
| SLTUNETModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-based2026.06 | 27.87 | 52.23 | — | — | — | — | — | |
| SIGNETModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 27.78 | 53.01 | 54.15 | — | — | — | — | |
| MSKA-SLTInput Modality=Keypoint2024.05 | 27.63 | 52.67 | 54.09 | 41.29 | 33.24 | — | — | |
| MSKA-SLTCategory=Gloss-Based, Extra Technique=3D, Pretrained; LM; Embedding, Params (M)=4042024.11 | 27.63 | — | — | — | — | — | — | |
| Ourspipeline_strategy=Sign2Text (end-to-end)2022.03 | 27.61 | 53.1 | 53.95 | 41.12 | 33.14 | — | — | |
| MMTLBLLM=true2023.08 | 27.61 | 53.1 | 53.95 | 41.12 | 33.14 | — | — | |
| MMTLBSetting=Gloss-based2024.04 | 27.61 | 53.1 | 53.95 | 41.12 | 33.14 | — | — | |
| MMTLBInput Modality=RGB2024.05 | 27.61 | 53.1 | 53.95 | 41.12 | 33.14 | — | — | |
| VL-MapperCategory=Gloss-Based, Extra Technique=Extra 3 Datasets; Pretrained, Params (M)=6102024.11 | 27.61 | — | — | — | — | — | — | |
| MMTLBCategory=Sign2Text2025.04 | 27.61 | 53.1 | 53.95 | 41.12 | 33.14 | — | — | |
| MMTLBModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-based2026.06 | 27.61 | 53.1 | 53.95 | — | — | — | — | |
| VL-Transfer2026.02 | 27.6 | 53.1 | 54 | 41.1 | 33.1 | — | — | |
| PGG-SLTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 27.53 | 52.85 | 53.84 | — | — | — | — | |
| Transformerstrategy=HATL2026.02 | 26.8 | 49.8 | 68 | 49.1 | 36.6 | — | — | |
| Two Stream-SLT2026.02 | 26.5 | 52 | 52.4 | 39.8 | 31.9 | — | — | |
| XmDAInput Modality=RGB2024.05 | 25.86 | 52.42 | — | — | — | — | — | |
| STN-LCUCategory=Gloss-Based, Extra Technique=Pretrained Embedding + Ensembling, 8 Models, Params (M)=2802024.11 | 25.59 | — | — | — | — | — | — | |
| MMSLTModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 25.47 | 48.58 | 48.73 | — | — | — | — | |
| SLTUNET2026.02 | 25.4 | 49.6 | — | — | — | — | — | |
| G2Ttype=gloss-to-text2021.09 | 25.35 | — | — | — | — | — | — | |
| SignLLMSetting=Gloss-free2024.04 | 25.25 | 47.23 | 46.88 | 36.59 | 29.91 | — | — | |
| SignLLMCategory=Gloss-Free, Extra Technique=LM + Pretraining; LLM, Params (M)=70122024.11 | 25.25 | — | — | — | — | — | — | |
| SignLLMModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 25.25 | 47.23 | 46.88 | — | — | — | — | |
| SignBERT+Input Modality=RGB, Refined Pose (R)=true2023.05 | 24.95 | 51.12 | 51.46 | 38.28 | 30.3 | — | — | |
| SignBERT+(+ R)Input Modality=RGB2024.05 | 24.95 | 51.12 | 51.46 | 38.28 | 30.3 | — | — | |
| Stochastic Transformer Networks with Linear Competing UnitsL=8, Bit-width=32 bit, Ensemble-Decoding=true2021.09 | 24.88 | — | — | — | — | — | — | |
| STMC-Transformer2026.02 | 24.7 | 48.7 | 50.3 | 37.6 | 29.8 | — | — | |
| S2G2T-STMC ensembletype=sign-to-gloss-to-text, ensemble=true2021.09 | 24.68 | — | — | — | — | — | — | |
| STMC-TCategory=Gloss-Based, Extra Technique=Pretrained; Transfer Learning + Ensembling, 8 Models, Params (M)=192*2024.11 | 24.68 | — | — | — | — | — | — | |
| Ourspipeline_strategy=Sign2Gloss2Text (two-stage)2022.03 | 24.63 | 50.23 | 50.36 | 37.5 | 29.69 | — | — | |
| VL-Transfer2026.02 | 24.6 | 50.2 | 50.4 | 37.5 | 29.7 | — | — | |
| Stochastic Transformer Networks with Linear Competing UnitsL=8, Bit-width=Reduced, Ensemble-Decoding=true2021.09 | 24.52 | — | — | — | — | — | — | |
| BN-TIN-Transf. + SignBTParadigm=S2T, SignBT=true2021.05 | 24.45 | 50.29 | 51.11 | 37.9 | 29.8 | — | — | |
| BN-TIN-Transf + BT*pipeline_strategy=Sign2Text (end-to-end), semi-supervised=true2022.03 | 24.45 | 50.29 | 51.11 | 37.9 | 29.8 | — | — | |
| SignBTInput Modality=RGB2023.05 | 24.45 | 50.29 | 51.11 | 37.9 | 29.8 | — | — | |
| BN-TIN-Transf.+SignBTSetting=Gloss-based2024.04 | 24.45 | 48.24 | 51.11 | 37.9 | 29.8 | — | — | |
| SignBTInput Modality=RGB2024.05 | 24.45 | 50.29 | 51.11 | 37.9 | 29.8 | — | — | |
| BN-TIN-Tran.+SignBTCategory=Gloss-Based, Extra Technique=Dataset; Pretrain; DataAug2024.11 | 24.45 | — | — | — | — | — | — | |
| SignBTCategory=Sign2Text2025.04 | 24.45 | 50.29 | 51.11 | 37.9 | 29.8 | — | — | |
| ConSLTTranslation Pipeline Setting=Cascading: Sign2Gloss2Text2022.04 | 24.27 | 47.52 | — | — | — | — | — | |
| ConSLTInput Modality=RGB2024.05 | 24.27 | 47.52 | — | — | — | — | — | |
| Stochastic Transformer Networks with Linear Competing UnitsL=4, Bit-width=Reduced, Ensemble-Decoding=true2021.09 | 24.23 | — | — | — | — | — | — | |
| STM-Net2026.02 | 24.1 | 48.2 | 47.6 | 36.4 | 29.2 | — | — | |
| STMC-TParadigm=S2T, multi-cue setting=true2021.05 | 24.09 | 48.24 | 47.6 | 36.43 | 29.18 | — | — | |
| STMC-Tpipeline_strategy=Sign2Text (end-to-end)2022.03 | 24.09 | 48.24 | 47.6 | 36.43 | 29.18 | — | — | |
| STMCLLM=false2023.08 | 24.09 | 48.24 | 47.6 | 36.43 | 29.18 | — | — | |
| STMC-TSetting=Gloss-based2024.04 | 24.09 | — | 47.6 | 36.43 | 29.18 | — | — | |
| STMC-TCategory=Sign2Text2025.04 | 24.09 | 48.24 | 47.6 | 36.43 | 29.18 | — | — | |
| STMCInput Modality=RGB2023.05 | 24.08 | 48.24 | 47.6 | 36.43 | 29.18 | — | — | |
| STMCInput Modality=RGB2024.05 | 24.08 | 48.24 | 47.6 | 36.43 | 29.18 | — | — | |
| Stochastic Transformer Networks with Linear Competing UnitsL=4, Bit-width=32 bit, Ensemble-Decoding=true2021.09 | 24.02 | — | — | — | — | — | — | |
| PETSetting=Gloss-based2024.04 | 24.02 | — | 49.54 | 37.19 | 29.3 | — | — | |
| BN-TIN-Transf.2 + BTParadigm=S2G2T, back-translation=true2021.05 | 23.51 | 49.53 | 49.33 | 36.43 | 28.66 | — | — | |
| BN-TIN-Transf + BT*pipeline_strategy=Sign2Gloss2Text (two-stage), semi-supervised=true2022.03 | 23.51 | 49.53 | 49.33 | 36.43 | 28.66 | — | — | |
| STN-SLTSetting=Gloss-based2024.04 | 23.23 | — | 49.12 | 36.29 | 28.34 | — | — | |
| STN-LCUCategory=Gloss-Based, Extra Technique=Pretrained Embedding, Params (M)=352024.11 | 23.23 | — | — | — | — | — | — | |
| STMC-Transformer2026.02 | 23.1 | 47.3 | 46.8 | 35 | 27.8 | — | — | |
| SimulSLT2026.02 | 22.9 | 49.2 | 47.8 | 35.3 | 27.9 | — | — | |
| HST-GNN2026.02 | 22.6 | — | 46.1 | 33.4 | 27.5 | — | — | |
| S2G2T-STMCtype=sign-to-gloss-to-text, ensemble=false2021.09 | 22.47 | — | — | — | — | — | — | |
| STMC-Transfpipeline_strategy=Sign2Gloss2Text (two-stage)2022.03 | 22.47 | 46.31 | 48.27 | 35.2 | 27.47 | — | — | |
| STMC-TransformerTranslation Pipeline Setting=Cascading: Sign2Gloss2Text2022.04 | 22.47 | 46.31 | — | — | — | — | — | |
| STMC-TCategory=Gloss-Based, Extra Technique=Pretrained; Transfer Learning, Params (M)=65*2024.11 | 22.47 | — | — | — | — | — | — | |
| Joint-SLRT2026.02 | 22.4 | — | 47.3 | 34.4 | 27.1 | — | — | |
| Joint-SLRTParadigm=S2T2021.05 | 22.38 | — | 47.26 | 34.4 | 27.05 | — | — | |
| SL-Transfpipeline_strategy=Sign2Text (end-to-end)2022.03 | 22.38 | — | 47.26 | 34.4 | 27.05 | — | — | |
| SL-TransInput Modality=RGB2023.05 | 22.38 | — | 47.26 | 34.4 | 27.05 | — | — | |
| Joint-SLRTLLM=false2023.08 | 22.38 | — | 47.26 | 34.4 | 27.05 | — | — | |
| SLRTSetting=Gloss-based2024.04 | 22.38 | 47.74 | 47.26 | 34.4 | 27.05 | — | — | |
| SL-TransInput Modality=RGB2024.05 | 22.38 | — | 47.26 | 34.4 | 27.05 | — | — | |
| SL-TransformerCategory=Gloss-Based, Extra Technique=Scratch, Params (M)=212024.11 | 22.38 | — | — | — | — | — | — | |
| Joint-SLRTCategory=Sign2Text2025.04 | 22.38 | — | 47.26 | 34.4 | 27.05 | — | — | |
| S2(G+T)type=sign-to-gloss-and-text2021.09 | 22.12 | — | — | — | — | — | — | |
| GFSLT-VLPSetting=Gloss-free2024.04 | 22.12 | 43.72 | 44.08 | 33.56 | 26.74 | — | — | |
| GFSLT-VIPCategory=Gloss-Free, Extra Technique=Pretrained; LM; Data Aug; Knowledge + Embedding; Transfer Learning, Params (M)=2172024.11 | 22.12 | — | — | — | — | — | — | |
| GFSLT-VLPModality (Pose)=true, Modality (RGB)=false, Approach=Gloss-free2026.06 | 22.12 | 43.72 | 44.08 | — | — | — | — | |
| SL-Transf.Paradigm=S2G2T2021.05 | 22.11 | — | 47.73 | 34.82 | 27.11 | — | — | |
| SL-Transfpipeline_strategy=Sign2Gloss2Text (two-stage)2022.03 | 22.11 | — | 47.73 | 34.82 | 27.11 | — | — | |
| SL-TransformerTranslation Pipeline Setting=Cascading: Sign2Gloss2Text2022.04 | 22.11 | — | — | — | — | — | — | |
| Joint-SLRTInput Modality=RGB2024.05 | 22.11 | — | 47.73 | 34.82 | 27.11 | — | — | |
| Signformer - FullCategory=Ours (Gloss-Free), Extra Technique=Scratch, Params (M)=3.882024.11 | 21.92 | — | — | — | — | — | — | |
| BN-TIN-Transf.2Paradigm=S2G2T, baseline=true2021.05 | 21.86 | 47.83 | 47.72 | 34.78 | 26.94 | — | — | |
| BN-TIN-Transfpipeline_strategy=Sign2Gloss2Text (two-stage)2022.03 | 21.86 | 47.83 | 47.72 | 34.78 | 26.94 | — | — |