# christophschuhmann/improved-aesthetic-predictor

CLIP+MLP Aesthetic Score Predictor

Repository: https://github.com/christophschuhmann/improved-aesthetic-predictor
Canonical: https://ross.abutalabs.com/products/improved-aesthetic-predictor
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2024-07-01T11:58:57+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1530, "days_push": 793, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1338, forks 114 (observed 2026-08-28T04:04:25.796009+00:00)

## What it is
A CLIP+MLP neural network that predicts how much people on average like an image, trained on AVA dataset ratings. It is widely used for filtering and ranking image datasets, notably LAION and Stable Diffusion training data.

## Use cases
- score the aesthetic quality of images
- filter image datasets for training diffusion models
- rank generated images by human preference
- curate LAION subsets by aesthetic score
- predict average human ratings for photos

## When to choose
- you need a lightweight aesthetic scorer built on CLIP embeddings
- you are curating large-scale image-text datasets
- you want a proven model used in Stable Diffusion data pipelines

## When to avoid
- you need fine-grained image quality metrics like sharpness or noise
- you cannot run CLIP embeddings
- you need video or audio aesthetic scoring

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, image-processing, data-science
- domain: machine-learning, artificial-intelligence, image-processing
- platform: python
- tags: clip, aesthetic-scoring, image-quality, laion, stable-diffusion, gpu

## Member repositories
- christophschuhmann/improved-aesthetic-predictor (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:25.796009+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:43:59.244456+00:00, confidence not recorded.
  - readme: https://github.com/christophschuhmann/improved-aesthetic-predictor (fetched 2026-08-28T04:04:25.796009+00:00, sha e822138ee888)
- Data as of 2026-08-30T08:39:29.467469+00:00.