Score FID
La Distance Fréchet Genèse Distance (FID) Score is a metric used to evaluate the quality of images generated by intelligence artificielle models, particularly generative adversarial networks (GANs). It quantifies how similar the generated images are to real images from a dataset, providing a numerical score that reflects the fidelity and diversity of the generated images.
Pour calculer le score FID, un modèle pré-entraîné réseau de neurones convolutionnels (CNN), often the Inception v3 model, is used to extract feature representations from both real and generated images. The key steps in calculating the FID Score involve:
- Extraction de caractéristiques: Les images sont passées dans le CNN pour obtenir des vecteurs de caractéristiques de haut niveau.
- Analyse statistique: The mean and covariance of these feature vectors are computed for both real and generated image sets.
- Calcul de la distance : The FID Score is then calculated using the Fréchet distance between the two Gaussian multivarié distributions définies par ces statistiques.
A lower FID Score indicates that the generated images are closer to the real images, suggesting higher quality. Conversely, a higher score implies that the generated images are less similar to the real ones. The FID Score is particularly useful because it takes into account both the quality (fidelity) and the diversity of the generated images, making it more reliable than simpler metrics comme une comparaison pixel par pixel.
En résumé, le score FID sert de référence importante dans le domaine de l'IA génération d'image, helping researchers and practitioners assess and improve their models.