https://github.com/urholaukkarinen/template-matching

GPU-accelerated template matching for Rust
https://github.com/urholaukkarinen/template-matching

computer-vision gpu image-recognition rust template-matching

Last synced: over 1 year ago
JSON representation

GPU-accelerated template matching for Rust

• Host: GitHub
• URL: https://github.com/urholaukkarinen/template-matching
• Owner: urholaukkarinen
• License: mit
• Created: 2023-07-19T19:08:37.000Z (almost 3 years ago)
• Default Branch: main
• Last Pushed: 2023-07-23T13:14:00.000Z (almost 3 years ago)
• Last Synced: 2024-04-29T00:23:21.997Z (about 2 years ago)
• Topics: computer-vision, gpu, image-recognition, rust, template-matching
• Language: Rust
• Homepage:
• Size: 65.4 KB
• Stars: 10
• Watchers: 1
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# template-matching

[![Latest version](https://img.shields.io/crates/v/template-matching.svg)](https://crates.io/crates/template-matching)

[![Documentation](https://docs.rs/template-matching/badge.svg)](https://docs.rs/template-matching)

![MIT](https://img.shields.io/badge/license-MIT-blue.svg)

GPU-accelerated template matching library for Rust. The crate is designed as a faster alternative to [imageproc::template_matching](https://docs.rs/imageproc/latest/imageproc/template_matching/index.html).

## Installation

```bash

[dependencies]

template-matching = { version = "0.2.0", features = ["image"] }

```

## Usage

```rust

use template_matching::{find_extremes, match_template, MatchTemplateMethod, TemplateMatcher};

fn main() {

    // Load images and convert them to f32 grayscale

    let input_image = image::load_from_memory(include_bytes!("input.png")).unwrap().to_luma32f();

    let template_image = image::load_from_memory(include_bytes!("template.png")).unwrap().to_luma32f();

    let result = match_template(&input_image, &template_image, MatchTemplateMethod::SumOfSquaredDifferences);

    // Or alternatively you can create the matcher first

    let mut matcher = TemplateMatcher::new();

    matcher.match_template(&input_image, &template_image, MatchTemplateMethod::SumOfSquaredDifferences);

    let result = matcher.wait_for_result().unwrap();

    // Calculate min & max values

    let extremes = find_extremes(&result);

}

```