NSFW API MRF

Upstream MR: https://git.pleroma.social/pleroma/pleroma/-/merge_requests/3471

Hide, delete, or mark sensitive NSFW content with artificial intelligence.

Requires a NSFW API server, configured like so:

  config :pleroma, Pleroma.Web.ActivityPub.MRF.NsfwMRF,
    url: "http://127.0.0.1:5000/",
    threshold: 0.8,
    mark_sensitive: true,
    unlist: false,
    reject: false

The NSFW API server must implement an HTTP endpoint like this:

  curl http://localhost:5000/?url=https://fedi.com/images/001.jpg

Returning a response like this:

  {"score", 0.314}

Where a score is 0-1, with 1 being definitely NSFW.

A good API server is here: https://github.com/EugenCepoi/nsfw_api You can run it with Docker with a one-liner:

  docker run -it -p 127.0.0.1:5000:5000/tcp --env PORT=5000 eugencepoi/nsfw_api:latest

Options:

• url: Base URL of the API server. Default: "http://127.0.0.1:5000/"
• threshold: Lowest score to take action on. Default: 0.7
• mark_sensitive: Mark sensitive all detected NSFW content? Default: true
• unlist: Unlist all detected NSFW content? Default: false
• reject: Reject all detected NSFW content (takes precedence)? Default: false