Lip Reading on Tulips1 (test)
0Model ParamsRAW
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RAWNumber of CNN Layers=0, RNN Architecture=None2023.04 | 0 | — | 57.45 | — | 1.5 | 1,048,576 | |
| Difference CameraNumber of CNN Layers=0, RNN Architecture=None2023.04 | 0 | 30 | — | 0.25 | — | 65,536 | |
| Event Camera*Number of CNN Layers=0, RNN Architecture=None2023.04 | 0 | 50 | — | 0.25 | — | 65,536 | |
| 1 CNN + SRNNNumber of CNN Layers=1, RNN Architecture=SRNN2023.04 | 451 | 50 | 90 | 2.38 | 76 | 4,096 | |
| 1 CNN + MGUNumber of CNN Layers=1, RNN Architecture=MGU2023.04 | 501 | 70 | 80 | 2.63 | 84 | 4,096 | |
| 1 CNN + PixelRNNNumber of CNN Layers=1, RNN Architecture=PixelRNN2023.04 | 501 | 80 | 100 | 2.75 | 88 | 4,096 | |
| 1 CNN + GRUNumber of CNN Layers=1, RNN Architecture=GRU2023.04 | 551 | 60 | 80 | 2.88 | 92 | 4,096 | |
| 1 CNN + genRNN (0,1)Number of CNN Layers=1, RNN Architecture=genRNN, Regime=(0,1)2023.04 | 553 | 60 | 80 | 3 | 96 | 4,096 | |
| 1 CNN + genRNN (-1,1)Number of CNN Layers=1, RNN Architecture=genRNN, Regime=(-1,1)2023.04 | 553 | 80 | 80 | 3 | 96 | 4,096 | |
| 1 CNN + LSTMNumber of CNN Layers=1, RNN Architecture=LSTM2023.04 | 601 | 50 | 90 | 3.38 | 108 | 4,096 | |
| 2 CNN + SRNNNumber of CNN Layers=2, RNN Architecture=SRNN2023.04 | 852 | 60 | 100 | 4.38 | 140 | 4,096 | |
| 2 CNN + MGUNumber of CNN Layers=2, RNN Architecture=MGU2023.04 | 902 | 60 | 90 | 4.63 | 148 | 4,096 | |
| 2 CNN + PixelRNNNumber of CNN Layers=2, RNN Architecture=PixelRNN2023.04 | 902 | 70 | 100 | 4.75 | 152 | 4,096 | |
| 2 CNN + GRUNumber of CNN Layers=2, RNN Architecture=GRU2023.04 | 952 | 70 | 100 | 4.88 | 156 | 4,096 | |
| 2 CNN + genRNN (0,1)Number of CNN Layers=2, RNN Architecture=genRNN, Regime=(0,1)2023.04 | 954 | 60 | 90 | 5 | 160 | 4,096 | |
| 2 CNN + genRNN (-1,1)Number of CNN Layers=2, RNN Architecture=genRNN, Regime=(-1,1)2023.04 | 954 | 60 | 90 | 5 | 160 | 4,096 | |
| 2 CNN + LSTMNumber of CNN Layers=2, RNN Architecture=LSTM2023.04 | 1,002 | 50 | 100 | 5.38 | 172 | 4,096 |