handtrack-rs

This repo provides a Rust API encapsulating the works of https://github.com/victordibia/handtracking. Basically this library encapsulates the required tensorflow interactions and provide a easy-to-use API for simply detecting hands.

This is very much in WIP and I have only tested this with single images. My aim here to basically be able to detect hands in real-time video just like the js version of this little library at https://github.com/victordibia/handtrack.js

Example

// Import the image.
let image = Image::from_file(project_dir).unwrap();

// Construct detection options.
let score_threshold = 0.7f32;
let max_hands = 1;
let detection_opts = DetectionOptions::new(max_hands, score_threshold);

// Run the detection.
let detection = detect(image, detection_opts).unwrap();

let detection_box = &detection[0];

As it can be seen from the example above, detect function requires an Image and DetectionOptions. Currently it is possible to specify desired maximum number of hands detected and score threshold for classifying an object as a hand.

Contribution

Although this is a small library it does have lots of missing features and contributions are more than welcome! As this is very early stage I do not have set contribution guidelines but I have some CI checks in place for just in case which are:

• clippy linting
• cargo fmt checking
• Cargo.toml linting (dependencies must be in alphabetical order etc.)
• cargo test check