# PeterWang512/CNNDetection

Code for the paper: CNN-generated images are surprisingly easy to spot... for now https://peterwang512.github.io/CNNDetection/

Repository: https://github.com/PeterWang512/CNNDetection
Canonical: https://ross.abutalabs.com/products/cnndetection
Language: Python
License: NOASSERTION
License Family: other
Last push: 2024-07-26T20:59:16+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": 2445, "days_push": 768, "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 1005, forks 185 (observed 2026-09-03T02:15:07.066687+00:00)

## What it is
A PyTorch research codebase with pretrained models for detecting CNN-generated (GAN/synthetic) images, from the CVPR 2020 paper 'CNN-generated images are surprisingly easy to spot...for now'. It includes demo scripts for single images and datasets, plus training and evaluation code and the paper's dataset.

## Use cases
- detect if an image was generated by a GAN
- evaluate a fake-image detector on a real/fake dataset
- train a CNN-generated image detector
- test robustness of synthetic image detection against JPEG and blur
- benchmark detection on StyleGAN3 outputs

## When to choose
- you need a pretrained detector for GAN-generated images
- you want to reproduce or extend CVPR 2020 synthetic image detection research
- you need AP/accuracy evaluation on a real-vs-fake image directory

## When to avoid
- you need detection of diffusion-model images rather than CNN/GAN-generated ones
- you want a production-ready API or service rather than research code
- you need a permissively licensed library - the license is non-standard

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, computer-vision, image-processing, testing
- domain: computer-vision, image-processing, deep-learning, artificial-intelligence
- platform: python, windows
- tags: deepfake-detection, gan-detection, synthetic-image-detection, pytorch, cvpr-2020, research-code, pretrained-models, linux, macos, gpu

## Member repositories
- PeterWang512/CNNDetection (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:07.066687+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 2026-08-30T07:12:44.565693+00:00, confidence not recorded.
  - readme: https://github.com/PeterWang512/CNNDetection (fetched 2026-09-03T02:15:07.066687+00:00, sha 9526d97f4063)
- Data as of 2026-08-30T08:39:29.467469+00:00.