# photosynthesis-team/piq

Measures and metrics for image2image tasks. PyTorch.

Repository: https://github.com/photosynthesis-team/piq
Canonical: https://ross.abutalabs.com/products/piq
Language: Python
License: Apache-2.0
License Family: permissive
Topics: image-quality-assessment, image-quality, image-to-image, metrics, measures, gan, generative-models, pytorch, python3, image-metrics, iqa, ssim, ms-ssim, mse, psnr, fid, kid, vif, brisque
Last push: 2024-05-12T15:30:07+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2449, "days_push": 843, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1574, forks 123 (observed 2026-08-28T04:05:05.733249+00:00)

## What it is
PyTorch Image Quality (PIQ) is a collection of measures and metrics for image quality assessment in image-to-image tasks, written in pure PyTorch. It provides a unified interface for metrics like SSIM, PSNR, FID, and BRISQUE, many of which can be used as differentiable loss functions.

## Use cases
- evaluate image quality of generated images
- compute SSIM or PSNR between two images in PyTorch
- use image quality metrics as loss functions for training
- measure FID or KID for GAN evaluation
- assess quality of super-resolution or denoising outputs
- benchmark image-to-image model performance on GPU

## When to choose
- you need a broad set of image quality metrics with a unified PyTorch API
- you want differentiable metrics usable as training losses
- you need GPU-accelerated IQA computation with input validation

## When to avoid
- you need metrics for TensorFlow, JAX, or non-PyTorch frameworks
- you only need a single simple metric and prefer a minimal dependency
- your project requires Python beyond the supported 3.7-3.10 range

## Facets
- artifact type: library
- maturity: stable
- function: image-processing, machine-learning, benchmarking, testing
- domain: image-processing, machine-learning, computer-vision, deep-learning
- platform: python, cross-platform
- tags: pytorch, image-quality-assessment, iqa, ssim, psnr, fid, kid, brisque, gan-metrics, loss-functions, gpu

## Member repositories
- photosynthesis-team/piq (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:05.733249+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 2026-08-30T03:57:51.469838+00:00, confidence not recorded.
  - readme: https://github.com/photosynthesis-team/piq (fetched 2026-08-28T04:05:05.733249+00:00, sha 9f2b0c794fdd)
  - registry_pypi: https://pypi.org/pypi/piq/json (fetched 2026-08-29T11:28:00.567008+00:00, sha 6634d0d5f66d)
- Data as of 2026-08-30T08:39:29.467469+00:00.