Video Prediction On Kth

평가 지표

Cond

FVD

LPIPS

PSNR

Params (M)

Pred

SSIM

Train

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름	Cond	FVD	LPIPS	PSNR	Params (M)	Pred	SSIM	Train	Paper Title	Repository
SVG-LP (from Grid-keypoints)	10	157.9	0.129	23.91	22.8	40	0.800	10	Stochastic Video Generation with a Learned Prior
SAVP-VAE	10	-	-	27.77	-	20	0.852	-	Stochastic Adversarial Video Prediction
Znet	10	-	-	27.58	-	20	0.817	-	Z-Order Recurrent Neural Networks for Video Prediction	-
MCnet + Residual	10	-	-	26.29	-	20	0.806	-	Decomposing Motion and Content for Natural Video Sequence Prediction
Conv-TT-LSTM	10	-	0.196	27.62	-	20	0.815	-	Convolutional Tensor-Train LSTM for Spatio-temporal Learning
DVG	-	-	-	-	-	-	-	-	Diverse Video Generation using a Gaussian Process Trigger
LMC	10	-	159.8	27.5	-	40	0.879	-	Video Prediction Recalling Long-term Motion Context via Memory Alignment Learning
SLAMP	10	228 ± 5	0.0795±0.0034	29.39±0.30	-	30	0.8646±0.0050	10	SLAMP: Stochastic Latent Appearance and Motion Prediction
SV2P time-invariant (from Grid-keypoints)	10	253.5	0.260	25.70	8.3	40	0.772	10	Stochastic Variational Video Prediction
PredRNN-V2	10	-	0.139	28.37	-	20	0.839	-	PredRNN: A Recurrent Neural Network for Spatiotemporal Predictive Learning
SRVP	10	222 ± 3	0.0736±0.0029	29.69±032	-	30	0.8697±0.0046	10	Stochastic Latent Residual Video Prediction
Struct-VRNN (from Grid-keypoints)	10	395.0	0.124	24.29	2.3	40	0.766	10	Unsupervised Learning of Object Structure and Dynamics from Videos
ConvLSTM	10	-	0.231	23.58	-	20	0.712	-	Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting
MSNET	10	-	-	27.08	-	20	0.876	-	Mutual Suppression Network for Video Prediction using Disentangled Features	-
SV2P time-invariant (from Grid-keypoints)	10	209.5	0.232	25.87	8.3	40	0.782	10	Stochastic Variational Video Prediction
SAVP-VAE (from Grid-keypoints)	10	145.7	0.116	26.00	7.3	40	0.806	10	Stochastic Adversarial Video Prediction
VarNet	10	-	-	28.48	-	20	0.843	-	VarNet: Exploring Variations for Unsupervised Video Prediction
DFN	10	-	-	27.26	-	20	0.794	-	Dynamic Filter Networks
MCnet	10	-	-	25.95	-	20	0.804	-	Decomposing Motion and Content for Natural Video Sequence Prediction
Grid-keypoints	10	144.2	0.092	27.11	2.0	40	0.837	10	Accurate Grid Keypoint Learning for Efficient Video Prediction

0 of 31 row(s) selected.