Image Generation on UTKFace (test)
0.425Intra-FIDCcGAN
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CcGANHyper-parameter selection strategy=SVDL, Label input mechanism=ILI, Image resolution=64x64, Training iterations=30,000, Batch size=2562020.11 | 0.425 | 1.725 | 1.298 | 7.452 | 3.382 | 0.142 | |
| CcGANHyper-parameter selection strategy=HVDL, Label input mechanism=ILI, Image resolution=64x64, Training iterations=30,000, Batch size=2562020.11 | 0.48 | 1.709 | 1.28 | 7.505 | 3.256 | 0.056 | |
| CcGANHyper-parameter selection strategy=SVDL, Label input mechanism=NLI, Image resolution=64x64, Training iterations=30,000, Batch size=2562020.11 | 0.547 | 1.753 | 1.326 | 10.739 | 3.307 | 0.087 | |
| CcGANHyper-parameter selection strategy=HVDL, Label input mechanism=NLI, Image resolution=64x64, Training iterations=30,000, Batch size=2562020.11 | 0.572 | 1.739 | 1.338 | 9.782 | 3.328 | 0.114 | |
| cGANLabel input mechanism=concat, Image resolution=64x64, Training iterations=30,000, Batch size=2562020.11 | 0.834 | 2.051 | 1.394 | 17.291 | 3.103 | 0.465 | |
| cGANNumber of classes (binning)=60 classes, Image resolution=64x64, Training iterations=30,000, Batch size=2562020.11 | 4.516 | 2.315 | 0.254 | 11.087 | 2.636 | 0.963 |