Skip to content
Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time


Build Status

select, download, and analyse media

mtriage is a command-line application to orchestrate complex scraping and analysis workflows. mtriage is developed at Forensic Architecture, and is intended for use by open source research agencies, journalists, and activists. To learn more about why we developed mtriage, you can read an overview of our reasons here.

getting started

First thing's first; follow the instructions to install mtriage:

Once installed, the best way to get started with mtriage is to work through the three tutorials:

latest update

Updates are posted irregularly; but you can get a sense of what's going on here by reading the latest update.

supported components

Below is a list of currently supported components. If you are interested in helping us to develop additional selectors and analysers, please consider joining the conversaton on Discord. We're accepting PRs for new components, but the internal documentation leaves a little bit wanting at the moment, so best to communicate with us directly on the #mtriage channel.


  • Youtube - search by query with optional date range (time uploaded), download video and metadata.
  • Twitter - search by query, download tweets and images.
  • Local - use media that already exists on your filesystem.


  • ConvertAudio - convert audio files between formats.
  • ExtractAudio - extract audio from a video.
  • ExtractTypes - extract elements that contain media with specified extensions.
  • Frames - extract frames from videos as images using ffmpeg.
  • ImageDedup - deduplicate images that are too similar using the imagededup module. (Good to use after using 'frames'.)
  • KerasPretrained - classify objects in images using Resnet50 trained on ImageNet.
  • Rank - generate a JSON file containing the rankings for videos classified with KerasPretrained.


framework to orchestrate the download and analysis of media




No packages published