FID in one dimension
Real images summarised by one Gaussian, generated ones by another. In 1-D the Fréchet distance is just the squared difference of means plus the squared difference of spreads, so the whole metric is visible at once.
Try this
Match the means exactly and leave the spreads different. FID stays stubbornly above zero — generating the right average is not generating the right variety.
FID = (μr − μg)² + (σr − σg)² · lower is better
Where this lab comes from
Image Generation Metrics: FID, IS, KID, LPIPS, and CLIPScoreHow do you score images with no ground truth? You compare distributions and use learned perceptual models: Inception Score, Fréchet Inception Distance, KID, LPIPS, and CLIPScore for prompt fidelity, on one running cat-generator example, with the equations, the catches, and torchmetrics.
More in metrics
- Precision, recall and the threshold — Edit the four counts, or drag a threshold and trace the PR and ROC curves.
- IoU and Dice, by dragging boxes — Two draggable boxes; intersection, union and both scores update live.
- Macro, micro and weighted, side by side — An editable C×C confusion matrix with per-class F1 drawn as bars.