List your company

ResearchFormerly WizWand

CelebV

Benchmarks

TaskDatasetSOTA resultTrendResults

Single-person talking-head generation

CelebV-HQ

3.875IQA

8

Cross-Reenactment

CelebV-HQ 69 (inference)

56.43FID

7

Self-Reenactment

CelebV-HQ 69 (inference)

30.84PSNR

7

Cross-identity reenactment

CelebV 30

79.1CSIM

7

Face Video Super-Resolution

CelebV-HQ

25.52PSNR

6

Audio-driven talking face generation

CelebV (randomly sampled 50 videos)

17.921FID

6

Video Generation

CelebV-HQ

34.01FID

5