V1.1.0

SCRFD v1.1.0 Release Notes 🎉

• Improved preprocessing image performance using opencv

Full Changelog: v1.0.0...v1.1.0

v1.0.0

SCRFD v1.0.0 Release Notes 🎉

We are excited to announce the first release of SCRFD, a high-performance Rust library for face detection. This release provides a robust and efficient foundation for face detection applications, leveraging ONNX Runtime for fast inference and supporting both synchronous and asynchronous workflows.

---

🚀 Features

• Face Detection:
  • Detect bounding boxes for faces in input images.
  • Optional keypoint detection for facial landmarks.
• Customizable Parameters:
  • Adjustable input size, confidence threshold, and IoU threshold.
• Optimized Processing:
  • Anchor-based detection with caching for anchor centers.
  • Preprocessing for efficient inference.
• Synchronous and Asynchronous Support:
  • Enable async workflows with the async feature flag.

---

📦 Installation

Add SCRFD to your Rust project:

[dependencies]
rusty_scrfd = { version = "0.1.0", features = ["async"] } # Optional async support

---

🛠️ Usage Examples

Synchronous Detection

use rusty_scrfd::SCRFD;
use image::open;
use ort::session::SessionBuilder;
use std::collections::HashMap;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let model_path = "path/to/scrfd_model.onnx";
    let session = SessionBuilder::new()?.with_model_from_file(model_path)?;

    let mut scrfd = SCRFD::new(session, (640, 640), 0.5, 0.4)?;

    let image = open("path/to/image.jpg")?.into_rgb8();
    let mut center_cache = HashMap::new();

    let (bboxes, keypoints) = scrfd.detect(&image, 5, "max", &mut center_cache)?;

    println!("Bounding boxes: {:?}", bboxes);
    if let Some(kps) = keypoints {
        println!("Keypoints: {:?}", kps);
    }

    Ok(())
}

Asynchronous Detection

use rusty_scrfd::SCRFDAsync;
use image::open;
use ort::session::SessionBuilder;
use std::collections::HashMap;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let model_path = "path/to/scrfd_model.onnx";
    let session = SessionBuilder::new()?.with_model_from_file(model_path)?;

    let scrfd = SCRFDAsync::new((640, 640), 0.5, 0.4, session)?;

    let image = open("path/to/image.jpg")?.into_rgb8();
    let mut center_cache = HashMap::new();

    let (bboxes, keypoints) = scrfd.detect(&image, 5, "max", &mut center_cache).await?;

    println!("Bounding boxes: {:?}", bboxes);
    if let Some(kps) = keypoints {
        println!("Keypoints: {:?}", kps);
    }

    Ok(())
}

---

🛠️ Bug Fixes and Improvements

• Initial release with a solid base for face detection and landmark prediction.
• Optimized for speed and flexibility.

---

🔖 Documentation

Full documentation is available in the README.

---

🛡️ License

This project is licensed under the MIT License.

---

❤️ Contributions

Contributions are welcome! Feel free to open issues or submit pull requests for any bugs, improvements, or features.

---

Enjoy using SCRFD! 🚀