Crate baklava 

Baklava is a rust wrapper around the InsightFace library to perform face comparison It returns a cosine and percentage similarity between the given image’s face against a target face.

The cosine similarity could be either positive or negative, depending on the orientation of the faces and how close they are to each other. Usually, a cosine similarity of 0.7 or higher is considered a good match.

The library can be used as simple as the example below.

use baklava::{InsightFace, Methodology};

let (cosine, percentage) = InsightFace::new("./Megatron", Some(3))
   .unwrap()
   .prepare_images(&[
       "./face1_test.png",
       "./face2_test.png",
   ]).unwrap()
   .prepare_target_image("./face1_test.png").unwrap()
   .compare_images(Methodology::Mean).unwrap();

To perform the comparison baklava required you to downlaod a model from the InsightFace repository that can be found at this link: https://github.com/HyperInspire/InspireFace?tab=readme-ov-file#resource-package-list

Modules

error

Structs

InsightFace

InsightFace is a struct which handle the internal pointers to compare two faces and returns the cosine value

Enums

Methodology

Methodology to use to compute get the cosine accross the selected image sources