Skip to content

ImagenHub

A library that standardises image model evaluation

Visit official site

What ImagenHub is

The library is described as a one-stop library for standardising inference and evaluation of conditional image generation models.

What you can do with it

  • Use seven defined generation tasks
  • Run models through a unified pipeline
  • Apply two human evaluation metrics

Application scenarios

  • Compare text-guided image generation
  • Assess mask-guided image editing
  • Review subject-driven image generation

Pros & cons

✓ What we like

  • A high-quality evaluation dataset is provided
  • About 30 models are evaluated in the study

! What to watch out for

  • Named scores come from human expert review
  • The material records 74 percent of models under a 0.5 overall score

FAQ

What metrics are designed?

Semantic consistency and perceptual quality are named.

How many models are evaluated?

The material states about 30 conditional image generation models.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review