Video Frame Interpolation on Vimeo90K (test)
36.69PSNRVFIformer (ours)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VFIformer (ours)strategy=with proposed strategy2023.12 | 36.69 | 0.9826 | |
| EMA-VFITraining Dataset=V, FLOPs (T)=0.91, Runtime (ms)=239.62024.07 | 36.64 | 0.9819 | |
| VFIMambaTraining Dataset=V+X, FLOPs (T)=0.94, Runtime (ms)=310.92024.07 | 36.64 | 0.9819 | |
| EMA-VFI2023.12 | 36.64 | 0.9819 | |
| AMT_oursparameters (millions)=36.74, runtime (seconds)=0.14, runtime_resolution=640 × 4802023.12 | 36.61 | 0.9828 | |
| Ours-Aug (DQBC)Params (M)=18.3, MACs (G)=229.9, Time (ms)=206, Test augmentation=true2023.04 | 36.57 | 0.9817 | |
| DQBC2023.12 | 36.57 | 0.9817 | |
| TTVFI2023.12 | 36.54 | 0.9819 | |
| VRTInputs=R2022.01 | 36.53 | 0.977 | |
| AMT-GTraining Dataset=V, FLOPs (T)=2.07, Runtime (ms)=403.72024.07 | 36.53 | 0.9817 | |
| VFIformerparameters (millions)=24.1, runtime (seconds)=1.0182022.11 | 36.5 | 0.9816 | |
| VFIformerParams (M)=24.2, MACs (G)=356.7, Time (ms)=7562023.04 | 36.5 | 0.9816 | |
| VFIformerRuntime (ms)=365, Param. (M)=24.1, Resolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 36.5 | 0.9816 | |
| AMTparameters (millions)=29.22, runtime (seconds)=0.06, runtime_resolution=640 × 4802023.12 | 36.5 | 0.9816 | |
| UPR-Net LARGEparameters (millions)=6.6, runtime (seconds)=0.0812022.11 | 36.42 | 0.9815 | |
| VFIformer-SmallRuntime (ms)=213, Param. (M)=17.0, Resolution=448 x 256, GPU=NVIDIA TITAN V, Window Size=4 x 4, Channel Number=1362022.05 | 36.38 | 0.9811 | |
| VFIformerstrategy=Baseline2023.12 | 36.38 | 0.9811 | |
| Ours (DQBC)Params (M)=18.3, MACs (G)=57.5, Time (ms)=54, Test augmentation=false2023.04 | 36.37 | 0.9812 | |
| AMT-LTraining Dataset=V, FLOPs (T)=0.58, Runtime (ms)=183.422024.07 | 36.35 | 0.9815 | |
| UPR-Net largeparameters (millions)=3.7, runtime (seconds)=0.0622022.11 | 36.28 | 0.981 | |
| FLAVRInputs=R2022.01 | 36.25 | 0.975 | |
| NCM (Ours-Large)2022.07 | 36.22 | 0.9807 | |
| IFRNet largeTime (s)=0.079, Params (M)=19.7, FLOPS (T)=0.792022.05 | 36.2 | 0.9808 | |
| IFRNet largeparameters (millions)=19.7, runtime (seconds)=0.0382022.11 | 36.2 | 0.9808 | |
| IFRNet-LTraining Dataset=V, FLOPs (T)=0.79, Runtime (ms)=115.32024.07 | 36.2 | 0.9808 | |
| IFRNet2023.12 | 36.2 | 0.9808 | |
| EBME-H*parameters (millions)=3.9, runtime (seconds)=0.0822022.11 | 36.19 | 0.9807 | |
| EBME2023.12 | 36.19 | 0.981 | |
| UPR-Net_oursparameters (millions)=2.64, runtime (seconds)=0.13, runtime_resolution=640 × 4802023.12 | 36.19 | 0.9806 | |
| ABMETime (s)=0.905, Params (M)=18.1, FLOPS (T)=1.32022.05 | 36.18 | 0.9805 | |
| ABMETraining Dataset=Vimeo-90K2022.02 | 36.18 | 0.981 | |
| ABME2022.07 | 36.18 | 0.9805 | |
| ABMEparameters (millions)=18.1, runtime (seconds)=0.2772022.11 | 36.18 | 0.9805 | |
| ABMEParams (M)=17.5, Time (ms)=2352023.04 | 36.18 | 0.9805 | |
| ABMERuntime (seconds)=0.22, #Parameters (millions)=18.12021.08 | 36.18 | 0.9805 | |
| ABMERuntime (ms)=158, Param. (M)=18.1, Resolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 36.18 | 0.9805 | |
| ABME2023.12 | 36.18 | 0.9805 | |
| RIFE-Large2022.07 | 36.13 | 0.98 | |
| SoftsplatTraining Dataset=V, FLOPs (T)=0.94, Runtime (ms)=266.42024.07 | 36.13 | 0.9805 | |
| SoftSplatTime (s)=0.195, Params (M)=12.2, FLOPS (T)=0.92022.05 | 36.1 | 0.97 | |
| SoftSplatLoss=L_Lap, Training Dataset=Vimeo-90K2022.02 | 36.1 | 0.97 | |
| SoftSplat2022.07 | 36.1 | 0.98 | |
| SoftSplat2022.11 | 36.1 | 0.97 | |
| SoftSplat-CResolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 36.1 | 0.97 | |
| RIFE-LargeRuntime (ms)=13, Param. (M)=21.7, Resolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 36.1 | 0.9801 | |
| SoftSplat2023.12 | 36.1 | 0.97 | |
| VFIMamba-STraining Dataset=V+X, FLOPs (T)=0.24, Runtime (ms)=128.02024.07 | 36.09 | 0.98 | |
| EMA-VFI-STraining Dataset=V, FLOPs (T)=0.20, Runtime (ms)=76.42024.07 | 36.07 | 0.9797 | |
| FILMLoss=L1, Training Dataset=Vimeo-90K2022.02 | 36.06 | 0.97 | |
| UPR-Netparameters (millions)=1.7, runtime (seconds)=0.0422022.11 | 36.03 | 0.9801 | |
| UPR-Netparameters (millions)=1.65, runtime (seconds)=0.05, runtime_resolution=640 × 4802023.12 | 36.02 | 0.98 | |
| NCM (Ours-Base)2022.07 | 35.88 | 0.9795 | |
| FILMLoss=Ls, Training Dataset=Vimeo-90K2022.02 | 35.87 | 0.968 | |
| SGM-VFITraining Dataset=V+X, FLOPs (T)=1.78, Runtime (ms)=942.92024.07 | 35.81 | 0.9793 | |
| IFRNetTime (s)=0.025, Params (M)=5, FLOPS (T)=0.212022.05 | 35.8 | 0.9794 | |
| IFRNetparameters (millions)=5.0, runtime (seconds)=0.0222022.11 | 35.8 | 0.9794 | |
| FILMLoss=LvGG, Training Dataset=Vimeo-90K2022.02 | 35.76 | 0.967 | |
| RIFETime (s)=0.026, Params (M)=9.8, FLOPS (T)=0.22022.05 | 35.62 | 0.978 | |
| RIFE2022.07 | 35.61 | 0.9779 | |
| RIFEparameters (millions)=9.8, runtime (seconds)=0.0122022.11 | 35.61 | 0.978 | |
| RIFEParams (M)=10.1, MACs (G)=11.7, Time (ms)=312023.04 | 35.61 | 0.9779 | |
| RIFETraining Dataset=V, FLOPs (T)=0.20, Runtime (ms)=35.22024.07 | 35.61 | 0.9779 | |
| IFRNet smallTime (s)=0.019, Params (M)=2.8, FLOPS (T)=0.122022.05 | 35.59 | 0.9786 | |
| InterpAny2023.12 | 35.51 | 0.9779 | |
| SoftSplatLoss=L_F, Training Dataset=Vimeo-90K2022.02 | 35.48 | 0.964 | |
| M2M-VFITraining Dataset=V, FLOPs (T)=0.26, Runtime (ms)=60.92024.07 | 35.47 | 0.9778 | |
| AdaCoFInputs=R2022.01 | 35.4 | 0.971 | |
| M2M-PWC_oursparameters (millions)=10.65, runtime (seconds)=0.04, runtime_resolution=640 × 4802023.12 | 35.37 | 0.9775 | |
| Bim-VFI_oursparameters (millions)=10.82, runtime (seconds)=0.06, runtime_resolution=640 × 4802023.12 | 35.31 | 0.9763 | |
| M2M-PWCparameters (millions)=7.61, runtime (seconds)=0.04, runtime_resolution=640 × 4802023.12 | 35.27 | 0.9771 | |
| CDFI fullTime (s)=0.38, Params (M)=5, FLOPS (T)=0.822022.05 | 35.17 | 0.964 | |
| CDFI fullparameters (millions)=5.0, runtime (seconds)=0.1722022.11 | 35.17 | 0.964 | |
| CDFIRuntime (ms)=60, Param. (M)=5.0, Resolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 35.17 | 0.964 | |
| QVIInputs=R+F2022.01 | 35.15 | 0.971 | |
| Bim-VFIparameters (millions)=6.56, runtime (seconds)=0.05, runtime_resolution=640 × 4802023.12 | 35.12 | 0.9753 | |
| XVFITraining Dataset=V, FLOPs (T)=0.37, Runtime (ms)=165.22024.07 | 35.09 | 0.9759 | |
| XVFI2022.07 | 35.07 | 0.9756 | |
| XVFIparameters (millions)=5.5, runtime (seconds)=0.0982022.11 | 35.07 | 0.9681 | |
| XVFIRuntime (ms)=96, Param. (M)=5.5, Resolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 35.07 | 0.976 | |
| BMBCRuntime (seconds)=0.77, #Parameters (million)=112020.07 | 35.01 | 0.9764 | |
| BMBCTime (s)=3.845, Params (M)=11, FLOPS (T)=2.52022.05 | 35.01 | 0.9764 | |
| BMBCTraining Dataset=Vimeo-90K2022.02 | 35.01 | 0.976 | |
| BMBC2022.07 | 35.01 | 0.9764 | |
| BMBCparameters (millions)=11.0, runtime (seconds)=0.8222022.11 | 35.01 | 0.9764 | |
| BMBCParams (M)=11.0, MACs (G)=174.2, Time (ms)=21682023.04 | 35.01 | 0.9764 | |
| BMBCRuntime (seconds)=0.77, #Parameters (millions)=11.02021.08 | 35.01 | 0.9764 | |
| BMBCRuntime (ms)=478, Param. (M)=11.0, Resolution=448 x 256, GPU=NVIDIA TITAN V2022.05 | 35.01 | 0.9764 | |
| BMBCInputs=R2022.01 | 34.76 | 0.965 | |
| DAINParameters (million)=24, Runtime (seconds)=0.132019.04 | 34.71 | 0.9756 | |
| DAINRuntime (seconds)=0.13, #Parameters (million)=242020.07 | 34.71 | 0.9756 | |
| DAINTime (s)=1.033, Params (M)=24, FLOPS (T)=5.512022.05 | 34.71 | 0.9756 | |
| DAIN2022.07 | 34.71 | 0.9756 | |
| DAINparameters (millions)=24.0, runtime (seconds)=0.1512022.11 | 34.71 | 0.9756 | |
| DAIN2023.04 | 34.71 | 0.9756 | |
| DAINTraining Dataset=V, FLOPs (T)=5.51, Runtime (ms)=897.82024.07 | 34.71 | 0.9756 | |
| DAINRuntime (seconds)=0.13, #Parameters (millions)=24.02021.08 | 34.71 | 0.9756 | |
| DAINTraining Dataset=Vimeo-90K2022.02 | 34.7 | 0.964 | |
| CAINTime (s)=0.069, Params (M)=42.8, FLOPS (T)=1.292022.05 | 34.65 | 0.973 | |
| CAINparameters (millions)=42.8, runtime (seconds)=0.0372022.11 | 34.65 | 0.973 | |
| CAINParams (M)=42.8, MACs (G)=43.5, Time (ms)=622023.04 | 34.65 | 0.973 |