Visual Speech Recognition on LRS3 v0.4 (test)
2.3WEROurs (raw A + V)
Evaluation Results
| Method | Links | |
|---|---|---|
| Ours (raw A + V)Training Data (Hours)=LRW (157) + LRS3v0.4 (438)2021.02 | 2.3 | |
| RNN-TTraining Data (Hours)=YT (31 000)2021.02 | 4.5 | |
| TM-seq2seqTraining Data (Hours)=MVLRS (730) + LRS2&3v0.4 (632)2021.02 | 7.2 | |
| RNN-TTraining Data (Hours)=YT (31 000)2021.02 | 33.6 | |
| Ours (V)Training Data (Hours)=LRW (157) + LRS3v0.4 (438)2021.02 | 43.3 | |
| Ours (V)Training Data (Hours)=LRS3v0.4 (438)2021.02 | 46.9 | |
| V2PTraining Data (Hours)=YT (3 886)2021.02 | 55.1 | |
| TM-seq2seqTraining Data (Hours)=MVLRS (730) + LRS2&3v0.4 (632)2021.02 | 58.9 | |
| KD + CTCTraining Data (Hours)=VC2clean (334) + LRS3v0.4 (438)2021.02 | 59.8 |