List your company

Video prediction on Kinetics-600 (test)

2.97FVDIV-VAE
-65.1112394.4369853.9851,313.5331Mar 9, 2020Dec 18, 2020Sep 29, 2021Jul 10, 2022Apr 21, 2023Jan 30, 2024Nov 10, 2024

Evaluation results

MethodLinks
IV-VAEParams=108M, FCR=4*8*8, Chn=162024.112.97——39.0296.850.0228—
CogX-VAEParams=215M, FCR=4*8*8, Chn=162024.113.17——38.3896.770.0287—
Causal VAEParams=127M, FCR=4*8*8, Chn=162024.113.28——38.0796.410.0289—
W.A.L.T-Lparams.=313M, steps=502023.123.3——————
IV-VAEParams=107M, FCR=4*8*8, Chn=82024.113.94——36.6695.20.036—
MAGVITv2params.=307M, steps=242023.124.3——————
Causal VAEParams=127M, FCR=4*8*8, Chn=82024.115.72——36.0894.790.0403—
SVD VAEParams=97M, FCR=1*8*8, Chn=42024.116.26——35.2693.630.0368—
IV-VAEParams=107M, FCR=4*8*8, Chn=42024.118.01——34.2992.810.0521—
MAGVITparams.=306M, steps=122023.129.9——————
OD-VAEParams=239M, FCR=4*8*8, Chn=42024.1110.69——33.8892.420.0549—
RINSteps=1000, GFLOPS=386, Param (M)=4112022.1210.8—17.7————
RINparams.=411M, steps=10002023.1210.8——————
Causal VAEParams=127M, FCR=4*8*8, Chn=42024.1110.81——33.7792.480.0631—
RINSteps=400, GFLOPS=386, Param (M)=4112022.1211.5—17.7————
Video Diffusionsampler=Langevin, steps=1282022.0416.2—15.64————
Video Diffusionparams.=1.1B, steps=2562023.1216.2——————
Video Diffusionreconstruction guidance=true, GFLOPS=4136, Param (M)=11002022.1216.6—15.64————
CV-VAEParams=182M, FCR=4*8*8, Chn=42024.1116.81——31.5890.660.0827—
Video Diffusionsampler=ancestral, steps=2562022.0418.6—15.39————
OS VAEParams=393M, FCR=4*8*8, Chn=42024.1119.05——34.3792.580.0816—
Transframer2022.0425.4——————
Transframer2022.1225.4——————
TrIVD-GAN-FP2022.1225.7—12.54————
TrIVD-GAN-FPsteps=12023.1225.7——————
TRIVD-GAN-FPResolution=64 x 642021.0325.74——————
TrIVD-GAN-FPpredicted frames=11, conditioned frames=5, frame skipping=false2020.0325.74—12.54————
TrIVD-GAN-FP2022.0425.74—12.54————
Phenakiparams.=227M, steps=482023.1236.4——————
Video VQ-VAEResolution=64 x 64, Evaluation Protocol=FVD* (Samples vs Reconstructions)2021.0354.3——————
CCVS2022.0455——————
Video VQ-VAEResolution=64 x 642021.0364.3——————
Video VQ-VAE2022.0464.3——————
Video VQ-VAE2022.1264.3——————
DVD-GAN-FP2022.0469.1——————
DVD-GAN-FP2022.1269.1——————
DVD-GAN-FPResolution=64 x 642021.0369.15——————
DVD-GAN-FPpredicted frames=11, conditioned frames=5, frame skipping=false2020.0369.15——————
Video VQ-VAEResolution=256 x 256, Evaluation Protocol=FVD* (Samples vs Reconstructions)2021.0382.45——————
Video VQ-VAEResolution=256 x 2562021.03129.85——————
Video TransformerResolution=64 x 642021.03170——————
Video Transformerpredicted frames=11, conditioned frames=5, frame skipping=false2020.03170——————
Video Transformer2022.04170——————
LVTnumber of codebooks (nc)=4, subscaling type=Single Frame2020.06224.732.14—————
Baseline2020.06271——————
Video Transformer2020.061,7051.19—————
MAGIType=AR, VAE=Omni2025.01——————11.5
MAGVITType=NAR, VAE=MAGVIT2025.01——————9.9
MAGVIT-v2Type=NAR, VAE=MAGVIT-v22025.01——————4.3
OmniType=AR, VAE=Omni2025.01——————32.9
PhenakiType=AR, VAE=VQVAE2025.01——————36.4
RINType=NAR2025.01——————10.8
Video DiffusionType=NAR2025.01——————16.2
ViVQVAEType=AR, VAE=VQVAE2025.01——————64.3