# facebookresearch/watermark-anything

Official implementation of the paper "Watermark Anything with Localized Messages"

Repository: https://github.com/facebookresearch/watermark-anything
Canonical: https://ross.abutalabs.com/products/watermark-anything
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: image, watermarking, image-watermarking
Archived: true
Last push: 2025-06-20T08:23:02+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 27, release rhythm 35, longevity 47
- inputs: {"age_days": 660, "days_push": 439, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1137, forks 55 (observed 2026-08-28T04:03:43.893169+00:00)

## What it is
Official PyTorch implementation and pretrained models for the paper 'Watermark Anything with Localized Messages', which embeds multiple localized watermarks into images. It provides inference notebooks, model weights under MIT license, and robustness against edits like moving watermarked objects.

## Use cases
- watermark images with invisible marks
- embed multiple localized watermarks in different image regions
- detect and decode watermarks from edited images
- verify image provenance or ownership
- research robust image watermarking with deep learning

## When to choose
- you need state-of-the-art localized or multi-region image watermarking
- you want MIT-licensed pretrained watermarking models
- you are doing research on watermark robustness and detection

## When to avoid
- you need video or audio watermarking
- you want a polished end-user GUI application rather than Python code
- you cannot run PyTorch on a GPU-capable machine

## Facets
- artifact type: library
- maturity: active
- function: image-processing, machine-learning, deep-learning, security
- domain: image-processing, computer-vision, artificial-intelligence, privacy
- platform: python, cross-platform
- tags: watermarking, image-watermarking, provenance, pytorch, pretrained-models, research-code, iclr-2025, gpu

## Member repositories
- facebookresearch/watermark-anything (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:43.893169+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 2026-08-30T06:36:14.584039+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/watermark-anything (fetched 2026-08-28T04:03:43.893169+00:00, sha 356f47ebbd70)
- Data as of 2026-08-30T08:39:29.467469+00:00.