Table 4 Quantitative results of different models on KTH action.
From: 3D long time spatiotemporal convolution for complex transfer sequence prediction
Methods | 10\(\rightarrow\)20 | 10\(\rightarrow\)40 | ||
|---|---|---|---|---|
SSIM\(\uparrow\) | PSNR\(\uparrow\) | SSIM\(\uparrow\) | PSNR\(\uparrow\) | |
ConvLSTM | 0.712 | 23.58 | 0.639 | 22.85 |
PredRNN | 0.839 | 27.55 | 0.703 | 24.16 |
E3D-LSTM | 0.879 | 29.31 | 0.810 | 27.24 |
WAM | 0.893 | 29.85 | 0.851 | 27.56 |
STAE | 0.899 | 29.91 | 0.859 | 27.87 |
STAM | 0.929 | 30.50 | 0.906 | 28.90 |
STMFANet | 0.893 | 29.85 | 0.851 | 27.56 |
SimVP | 0.905 | 33.72 | 0.886 | 32.93 |
SwinLSTM | 0.903 | 34.34 | 0.879 | 33.15 |
VPTR-FAR | 0.859 | 26.13 | - | - |
VPTR-PAR | 0.836 | 25.40 | - | - |
VPTR-PAR | 0.879 | 26.96 | - | - |
Ours | 0.920 | 29.95 | 0.912 | 28.94 |
