Notices
Last updated 10 September 2026
dismogs is built on open-source work. This page credits the parts that ask to be credited, and names every model that looks at a photograph before it is stored.
The photo filter
A photograph added to dismogs is checked, on our own server, by four open models before anything is kept. Nothing is sent to another company, and none of these does face recognition: they count faces, find one box, estimate an age, and read what is written on the picture. What that check is for is on the privacy page.
Ultra-Light-Fast-Generic-Face-Detector (RFB-320) — MIT, Copyright (c) 2019 linzai. Finds the faces, and how many there are.
abhilash88/age-gender-prediction — Apache License 2.0. Estimates an age from the face. We read only the age; the model also predicts a sex and we discard it unread.
AdamCodd/vit-base-nsfw-detector — Apache License 2.0. Judges whether the picture is explicit.
Tesseract OCR — Apache License 2.0. Reads text on the picture, so a phone number or a handle written across a photograph is caught before it is published.
Everything else
jsQR — Apache License 2.0. Looks for QR codes in an uploaded picture.
sharp — Apache License 2.0. Decodes and resizes every image, and strips its metadata.
onnxruntime — MIT, Copyright (c) Microsoft Corporation. Runs the models above.
Transformers.js — Apache License 2.0. Loads and pre-processes two of them.
Archivo and IBM Plex Mono — SIL Open Font License 1.1. The type, everywhere — on the site, in the Activity, and in every drawn card.
Twemoji — CC BY 4.0, Copyright Twitter, Inc and other contributors. The emoji drawn into card images.
dismogs also runs on Next.js, React and Tailwind CSS (MIT), and on Supabase and Vercel. A full dependency tree is longer than this page and available on request.
Corrections
If your work is here and credited wrongly, or is missing, tell us and we will fix it: hello@dismogs.com.