# cszn/BSRGAN

Designing a Practical Degradation Model for Deep Blind Image Super-Resolution (ICCV, 2021) (PyTorch) - We released the training code!

Repository: https://github.com/cszn/BSRGAN
Canonical: https://ross.abutalabs.com/products/bsrgan
Language: Python
License: Apache-2.0
License Family: permissive
Topics: blind-image-super-resolution, super-resolution, real-image-super-resolution, realsr
Last push: 2024-04-12T18:35:45+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": 2050, "days_push": 873, "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 1386, forks 191 (observed 2026-08-28T04:04:34.956144+00:00)

## What it is
BSRGAN is a PyTorch implementation of a practical degradation model for deep blind image super-resolution, presented at ICCV 2021. It provides trained models and a degradation pipeline that synthesizes realistic low-quality images for training super-resolution networks.

## Use cases
- upscale real-world low-quality photos with deep learning
- generate realistic degraded training data for super-resolution models
- train a blind super-resolution network in PyTorch
- enhance resolution of images with unknown degradations
- apply BSRGAN pretrained models at scale factors 2 and 4

## When to choose
- you need to super-resolve real-world images with mixed, unknown degradations
- you want a realistic degradation pipeline to train blind SR models
- you want pretrained PyTorch models with available training code

## When to avoid
- you only need classical bicubic/Lanczos upscaling without deep learning
- you need lightweight on-device super-resolution rather than GPU-based research models
- you want a maintained product-grade tool rather than research code

## Facets
- artifact type: library
- maturity: stable
- function: image-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, deep-learning, machine-learning
- platform: python, cross-platform
- tags: super-resolution, blind-super-resolution, pytorch, degradation-model, image-restoration, research-code, iccv-2021, gpu

## Member repositories
- cszn/BSRGAN (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:34.956144+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 2026-08-30T04:39:53.735846+00:00, confidence not recorded.
  - readme: https://github.com/cszn/BSRGAN (fetched 2026-08-28T04:04:34.956144+00:00, sha 230fb4c5edbf)
- Data as of 2026-08-30T08:39:29.467469+00:00.