Adversarial Video Generation on Complex Datasets

15 Jul 2019 Aidan Clark Jeff Donahue Karen Simonyan

Generative models of natural images have progressed towards high fidelity samples by the strong leveraging of scale. We attempt to carry this success to the field of video modeling by showing that large Generative Adversarial Networks trained on the complex Kinetics-600 dataset are able to produce video samples of substantially higher complexity and fidelity than previous work... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Video Generation Kinetics-600 12 frames, 128x128 DVD-GAN FID 2.16 # 1
Video Generation Kinetics-600 12 frames, 64x64 DVD-GAN FID 0.91 # 1
Inception Score 129.9 # 1
Video Generation Kinetics-600 48 frames, 64x64 DVD-GAN FID 12.92 # 1
Inception Score 219.05 # 1

Methods used in the Paper