# taesungp/contrastive-unpaired-translation

Contrastive unpaired image-to-image translation, faster and lighter training than cyclegan (ECCV 2020, in PyTorch)

Repository: https://github.com/taesungp/contrastive-unpaired-translation
Canonical: https://ross.abutalabs.com/products/contrastive-unpaired-translation
Homepage: https://taesung.me/ContrastiveUnpairedTranslation/
Language: Python
License: NOASSERTION
License Family: other
Topics: pytorch, computervision, deeplearning, cyclegan, image-generation, computer-vision, computer-graphics, image-manipulation, gans, generative-adversarial-network
Last push: 2023-09-05T16:46:25+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2234, "days_push": 1093, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2543, forks 451 (observed 2026-08-28T04:06:59.487598+00:00)

## What it is
A PyTorch implementation of Contrastive Unpaired Translation (CUT), an ECCV 2020 method for unpaired image-to-image translation using patchwise contrastive learning and adversarial training. It trains faster and with less memory than CycleGAN and supports single-image domain translation.

## Use cases
- translate images between domains without paired training data
- train an image-to-image translation model faster than CycleGAN
- perform style transfer between unpaired image collections
- train a translation model from a single image per domain
- convert photos to paintings or between visual styles with GANs

## When to choose
- you need unpaired image-to-image translation with lower training cost than CycleGAN
- you want a research-grade PyTorch codebase for contrastive learning in image synthesis
- you only have a single image per domain to train on

## When to avoid
- you need paired image-to-image translation like pix2pix
- you need a production-ready, actively maintained library with long-term support
- you lack a GPU and need fast training

## Facets
- artifact type: library
- maturity: maintenance
- function: image-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, deep-learning, machine-learning
- platform: python
- tags: pytorch, gan, image-translation, contrastive-learning, cyclegan, research-code, eccv-2020, linux, macos, gpu

## Member repositories
- taesungp/contrastive-unpaired-translation (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:59.487598+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 2026-08-30T02:25:06.145951+00:00, confidence not recorded.
  - readme: https://github.com/taesungp/contrastive-unpaired-translation (fetched 2026-08-28T04:06:59.487598+00:00, sha 13fe62503e33)
  - homepage: https://taesung.me/ContrastiveUnpairedTranslation/ (fetched 2026-08-29T10:07:05.782583+00:00, sha 79e4ea4904c9)
- Data as of 2026-08-30T08:39:29.467469+00:00.