The Role of ImageNet Classes in Fréchet Inception Distance

No Thumbnail Available

Access rights

openAccess

URL

Journal Title

Journal ISSN

Volume Title

A4 Artikkeli konferenssijulkaisussa

Date

2023-05-01

Major/Subject

Mcode

Degree programme

Language

en

Pages

26

Series

Eleventh International Conference on Learning Representations (ICLR 2023)

Abstract

Fréchet Inception Distance (FID) is the primary metric for ranking models in data-driven generative modeling. While remarkably successful, the metric is known to sometimes disagree with human judgement. We investigate a root cause of these discrepancies, and visualize what FID "looks at" in generated images. We show that the feature space that FID is (typically) computed in is so close to the ImageNet classifications that aligning the histograms of Top-N classifications between sets of generated and real images can reduce FID substantially -- without actually improving the quality of results. Thus, we conclude that FID is prone to intentional or accidental distortions. As a practical example of an accidental distortion, we discuss a case where an ImageNet pre-trained FastGAN achieves a FID comparable to StyleGAN2, while being worse in terms of human evaluation.

Description

| openaire: EC/H2020/866435/EU//PIPE

Keywords

Other note

Citation

Kynkäänniemi, T, Karras, T, Aittala, M, Aila, T & Lehtinen, J 2023, The Role of ImageNet Classes in Fréchet Inception Distance . in 11th International Conference on Learning Representations (ICLR 2023) . Curran Associates Inc., International Conference on Learning Representations, Kigali, Rwanda, 01/05/2023 . < https://arxiv.org/abs/2203.06026 >