cff-version: 1.2.0
message: "If you use SynthScope in your research, please cite it."
title: "SynthScope: A Unified Library and Benchmark for AI-Generated Image and Video Detection"
abstract: >-
  SynthScope re-implements a range of state-of-the-art AI-generated (synthetic) image and video
  detectors behind a single configuration system, training loop, datasets, transforms, and
  metrics, so that different methods can be trained and benchmarked under identical conditions,
  with results reported per generator/source.
type: software
authors:
  # TODO(maintainer): verify the author name(s) and affiliation, and add ORCID(s).
  - family-names: "Salehi Dehnavi"
    given-names: "Seyed"
    email: "s.salehidehnavi@uva.nl"
    affiliation: "University of Amsterdam"
repository-code: "https://github.com/SMSD75/SynthScope"
url: "https://github.com/SMSD75/SynthScope"
license: Apache-2.0
version: "0.1.0"
keywords:
  - AI-generated image detection
  - synthetic media detection
  - deepfake detection
  - benchmark