R/RcppExports.R

In lime: Local Interpretable Model-Agnostic Explanations

# Generated by using Rcpp::compileAttributes() -> do not edit by hand
# Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393

rowSumsSq <- function(sparse_matrix) {
    .Call('_lime_rowSumsSq', PACKAGE = 'lime', sparse_matrix)
}

get_index_permutations <- function(original_document, number_permutations) {
    .Call('_lime_get_index_permutations', PACKAGE = 'lime', original_document, number_permutations)
}

#' @title Segment image into superpixels
#'
#' @description This is an implementation of the SLIC superpixel algorithm for
#' segmenting images into connected similar patches. It is used by lime for
#' permuting image input but exported so that others might use it for other
#' things
#'
#' @param L,a,b Raw matrices giving the L, a, and b component of each pixel in
#' the image to segment. The dimensions of all matrices must match.
#'
#' @param n_sp The number of superpixels to segment the image into
#'
#' @param weight A numeric giving the tradeoff between spatial and colour
#' distance. Higher values give more compact and heterogeneous superpixels,
#' while lower values will give superpixels of more irregular shape but with a
#' more homogeneous colour. Good values to start with is 10-20.
#'
#' @param n_iter The number of iterations to run the algorithm for. The authors
#' suggest 10 and increasing it doesn't add much.
#'
#' @return An integer matrix of the same dimensions as `L`, `a`, and `b`,
#' indexing each pixel into its corresponding superpixel
#'
#' @keywords internal
#' @export
slic <- function(L, a, b, n_sp, weight, n_iter) {
    .Call('_lime_slic', PACKAGE = 'lime', L, a, b, n_sp, weight, n_iter)
}