https://github.com/podgorskiy/kmeans

Simple, one header drop in templated library from kmeans
https://github.com/podgorskiy/kmeans

Last synced: 8 months ago
JSON representation

Simple, one header drop in templated library from kmeans

• Host: GitHub
• URL: https://github.com/podgorskiy/kmeans
• Owner: podgorskiy
• Created: 2020-11-29T02:59:21.000Z (almost 5 years ago)
• Default Branch: main
• Last Pushed: 2020-11-29T03:17:46.000Z (almost 5 years ago)
• Last Synced: 2025-01-11T05:21:44.895Z (10 months ago)
• Language: C++
• Size: 48.8 KB
• Stars: 2
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# kmeans

Simple, one-header, drop-in, templated library for kmeans

![jpeg](imgs/296059.jpg)

![jpeg](imgs/kmeans4_rgb_2__.png)

Works out of the box with scalars and vector types from OpenCV and GLM, such as glm::vec2, glm::vec3, cv::Vec3b, etc (Euclidean metric is assumed).

For everything else, provide specialization of:

```cpp

namespace kmeans

{

	// Provide your own specialization for your data type

	// Already have default ones for scalars and vector types as in cv::Vec<> and glm::vec_t<>

	template

	class DataTypeImpl

	{

	public:

		typedef float DistanceType;

		static T GenerateRandomSample(const std::pair& bounds);

		static std::pair GetBounds(const T* first, const T* end);

		static DistanceType ComputeDistance(const T& a, const T& b);

		static T ComputeMean(const T* first, const T* end);

	};

}

```

Where you need to 

 * specify `DistanceType` (typically float)

 * specify function for random sampling `GenerateRandomSample`, given the bounds

 * specify function for computing bounds `GetBounds`, given the sample list

 * specify function for computing distance `ComputeDistance`, given the two samples which can be non euclidean to.

 * specify function for computing mean of given sample list `ComputeMean`.