TrustMark — Rust

An implementation of TrustMark watermarking for the Content Authenticity Initiative (CAI) in Rust, as described in:

TrustMark - Universal Watermarking for Arbitrary Resolution Images

https://arxiv.org/abs/2311.18297

Tu Bui¹, Shruti Agarwal², John Collomosse^1,2

¹DECaDE Centre for the Decentralized Digital Economy, University of Surrey, UK.
²Adobe Research, San Jose CA.

This is a re-implementation of the trustmark Python library.

Quick start

cargo add trustmark

Download models

From the workspace root, run:

$ cargo xtask fetch-models

Run the CLI

From the workspace root, run:

$ cargo run --release -p trustmark-cli -- -m ./models encode -i ./images/bfly_rgba.png -o ./images/encoded.png
$ cargo run --release -p trustmark-cli -- -m ./models decode -i ./images/encoded.png

Use the library

use trustmark::{Trustmark, Version, Variant};

let tm = Trustmark::new("./models", Variant::Q, Version::Bch5).unwrap();
let input = image::open("../images/ghost.png").unwrap();
let output = tm.encode("0010101".to_owned(), input, 0.95);

Running the benchmarks

Rust benchmarks

To run the Rust benchmarks, run:

cargo bench

Python benchmarks

To run the Python benchmarks, run the following from the workspace root:

benches/load.sh && benches/encode.sh

Differences from the Python version

This crate implements a subset of the functionality of the Python version. Open an issue if there's something in the Python version that would be useful in this crate!