https://github.com/kayagokalp/handtrack-rs

A library for prototyping hand detection and tracking (bounding box) for rust
https://github.com/kayagokalp/handtrack-rs

artificial-intelligence handtracking neural-networks rust tensorflow

Last synced: 25 days ago
JSON representation

A library for prototyping hand detection and tracking (bounding box) for rust

• Host: GitHub
• URL: https://github.com/kayagokalp/handtrack-rs
• Owner: kayagokalp
• License: mit
• Created: 2022-12-25T20:28:22.000Z (almost 2 years ago)
• Default Branch: master
• Last Pushed: 2023-01-10T21:11:44.000Z (almost 2 years ago)
• Last Synced: 2024-09-17T12:10:04.888Z (about 2 months ago)
• Topics: artificial-intelligence, handtracking, neural-networks, rust, tensorflow
• Language: Rust
• Homepage:
• Size: 16.2 MB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 5
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.MD

Awesome Lists containing this project

README

        
# 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

```rust

// 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