Struct KmeansOptions 

pub struct KmeansOptions { /* private fields */ }

The various options for k-means quantization.

This struct has a builder API. See the docs for each of the following functions for more details:

• sampling_factor
• max_samples
• batch_size
• seed

Examples

KmeansOptions::new()
    .sampling_factor(0.5)
    .max_samples(512 * 512 / 2)
    .batch_size(2048)
    .seed(42);

Implementations

impl KmeansOptions

pub const fn new() -> Self

Create a new KmeansOptions with default options.

pub const fn sampling_factor(self, sampling_factor: f32) -> Self

Sets the proportion of the input to sample.

This is typically in the range 0.0..=1.0. It can be above 1.0, but this may not give noticeably better results. You can also set this to f32::INFINITY to instead use max_samples to limit the number of samples independent of the input size.

The default sampling factor is 1.0.

pub const fn max_samples(self, max_samples: u32) -> Self

Sets the maximum number of pixels to sample from the input.

The number of samples determined by the sampling_factor is proportional to the input size. However, samples after a certain point will likely not affect the results in a significant way, since the k-means quantizition incorporates a learning rate that increasingly diminishes the impact of new samples. So, you can use this option to limit the number samples in the case of a large input.

Also, you can set sampling_factor to f32::INFINITY, which means that this setting will determine the exact number of samples that will occur.

The default maximum samples is 262144.

pub const fn batch_size(self, batch_size: u32) -> Self

Sets the number of samples to batch together each iteration. This option is only used by the parallel versions of the k-means quantization functions.

Increasing the batch size reduces the running time but with dimishing returns. Smaller batch sizes are more accurate but are slower to run.

The default batch size is 4096.

pub const fn seed(self, seed: u64) -> Self

Sets the seed number used for the random number generators.

The default seed is 0.

pub const fn get_sampling_factor(&self) -> f32

Returns the current sampling factor.

See sampling_factor for more information.

pub const fn get_max_samples(&self) -> u32

Returns the current maximum number of samples.

See max_samples for more information.

pub const fn get_batch_size(&self) -> u32

Returns the current batch size.

See batch_size for more information.

pub const fn get_seed(&self) -> u64

Returns the current seed number.

See seed for more information.

Trait Implementations

impl Clone for KmeansOptions

fn clone(&self) -> KmeansOptions

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for KmeansOptions

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for KmeansOptions

fn default() -> Self

Returns the “default value” for a type.

impl From<KmeansOptions> for QuantizeMethod

Available on crate feature kmeans only.

fn from(options: KmeansOptions) -> Self

Converts to this type from the input type.

impl Hash for KmeansOptions

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for KmeansOptions

fn eq(&self, other: &KmeansOptions) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for KmeansOptions

impl Eq for KmeansOptions

impl StructuralPartialEq for KmeansOptions