Rethinking FID Through the Geometry of the Reference Dataset

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lee, Yunghee, Pak, Byeonghyun
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914612159447040
author Lee, Yunghee
Pak, Byeonghyun
author_facet Lee, Yunghee
Pak, Byeonghyun
contents Fréchet Inception Distance (FID) is widely used to evaluate image generators, yet lower FID does not always correspond to better sample quality. We show that this mismatch depends in part on the geometry of the reference dataset. In a controlled study across six datasets, distributional density and effective rank significantly explain how FID changes as sample quality improves. Concentrated datasets tend to yield more favorable FID trends, whereas more dispersed datasets can make FID worsen despite better samples. Attribution to precision and recall and ablations with alternative feature spaces and distances support the same conclusion. These results suggest that distributional metrics should be interpreted together with the geometry of the reference dataset for more reliable benchmarking.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29335
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Rethinking FID Through the Geometry of the Reference Dataset
Lee, Yunghee
Pak, Byeonghyun
Computer Vision and Pattern Recognition
Artificial Intelligence
Fréchet Inception Distance (FID) is widely used to evaluate image generators, yet lower FID does not always correspond to better sample quality. We show that this mismatch depends in part on the geometry of the reference dataset. In a controlled study across six datasets, distributional density and effective rank significantly explain how FID changes as sample quality improves. Concentrated datasets tend to yield more favorable FID trends, whereas more dispersed datasets can make FID worsen despite better samples. Attribution to precision and recall and ablations with alternative feature spaces and distances support the same conclusion. These results suggest that distributional metrics should be interpreted together with the geometry of the reference dataset for more reliable benchmarking.
title Rethinking FID Through the Geometry of the Reference Dataset
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2605.29335