Como agrupar papéis de parede / fotos de acordo com suas cores?

Question

Eu tenho vários espaços de trabalho e gostaria que eles fossem distinguíveis pelos papéis de parede.

Por outro lado, gostaria de ter alguma forma de rotação de papel de parede em cada espaço de trabalho.

Existe algum software pronto, que pode levar meus papéis de parede como entrada, e agrupá-los em número especificado de grupos por sua cor?

Bônus extra se funcionasse na linha de comando (para que eu possa automatizá-lo).

    

Answer 1

Eu escrevi o seguinte script com R:

#!/usr/bin/Rscript

makeSureInstalled<-function(package) {if (length(grep(paste("^",package,"$",sep=""),noquote(installed.packages())[,1]))==0) install.packages(package)
                                      library(package=package,character.only=TRUE)
}

makeSureInstalled("argparse")
makeSureInstalled("foreach")
makeSureInstalled("doMC")

arghandling<-function()
  # Reads command line arguments, verifies the input
{
  parser <- ArgumentParser(description="Reads all files in the specified directory (not-recursively) treats them as jpegs, and groups them by mean color.",
                           epilog="Example: ./wallpapers-grouping.R --cluster-count 9 --out-prefix 'cluster '")
  parser$add_argument("--cluster-count","-n", type="integer", dest="clustercount",
                      help="Number of groups to do. Must be (much) smaller than number of individual photos.")
  parser$add_argument("--feature-set", "-f", dest="mode", type="integer", 
                      help="Before running the actual clustering, the program pre-processes image and extracts features. There are 2 feature sets supported: 1 - one pixel (3 numbers for each colour). 2 - 4 pixel (image is divided into equal 4 parts, and each part is averaged to one pixel. 4x3=12 numbers in total.)")
  parser$add_argument("--out-prefix","-o", type="character", default='cluster ', dest='prefix',
                      help="The prefix appended to each grouped picture before grouping. Can be a directory name (absolute or relative) or name prefix.")
  parser$add_argument("--cpu-count",  type="integer", dest="cpu",
                      help="How many cores to use. Defaults to all available cores.")
  parser$add_argument("input", type="character", 
                      help="Input directory. Defaults to current directory. ")

  args<-commandArgs(trailingOnly = TRUE)

  if (length(args)==0)
  {
    parser$print_help()
    quit()
  } else
  {
    cat("Welcome to the wallpapers-grouping.R program.\n")
    args=parser$parse_args(args)
  }


  if (is.null(args$clustercount))
  {
    cat(paste0("Error! You must specify number of clusters with --cluster-count or -n argument.\n"))
    quit()    
  }

  if (is.null(args$input))
  {
    args$input<-cwd()
  } else {
    if(!file.exists(args$input))
    {
      cat(paste0("Cannot find the input directory (", args$input,")\n"))
      quit()
    }
  }

  photos=dir(path=args$input)
  cat(paste0(length(photos)," files found in ",args$input,".\n"))

  if (length(photos)<=args$clustercount)
  {
    cat(paste0("Number of clusters must be (much) smaller than number of photos.\n"))
    quit()
  }

  registerDoMC()
  howManyCores<-getDoParWorkers()
  if(is.null(args$cpu) || args$cpu==howManyCores)
  {
    cat(paste0(howManyCores," cores detected.\n"))
  } else {
    cat(paste0(howManyCores," cores detected, but will use ",args$cpu," independent threads.\n"))  
    howManyCores<-args$cpu
  }
  registerDoMC(cores=howManyCores)

  if(is.null(args$mode))
  {
    args$mode<-1
  }

  args$mode<-switch(args$mode,1,2)

  list(args=args,photos=photos)
}

ah<-arghandling()

photos<-ah$photos
args<-ah$args


makeSureInstalled('jpeg')
makeSureInstalled('plyr')

processphoto<-function(filename,mode=1) {
  photo=readJPEG(filename)
  if(mode==1)
  {
    apply(photo,3,mean)
  }else  {
    dims<-dim(obraz)[1:2]
    gr=round(dims/2)
    o11<-photo[1:gr[[1]],1:gr[[2]],]
    o12<-photo[(gr[[1]]+1):dims[[1]],1:gr[[2]],]
    o21<-photo[1:gr[[1]],(gr[[2]]+1):dims[[2]],]
    o22<-photo[(gr[[1]]+1):dims[[1]],(gr[[2]]+1):dims[[2]],]
    o11<-apply(o11,3,mean)
    o12<-apply(o12,3,mean)
    o21<-apply(o21,3,mean)
    o22<-apply(o22,3,mean)

    c(o11,o12,o21,o22)
  }
}

cat(paste0('Extracting features from all ',length(photos),' pictures...'))
results<-aaply(photos,1,processphoto,.parallel=TRUE,mode=args$mode)

cat(paste0('Grouping them into ',length(args$clustercount),' groups using k-means...'))
clusters<-kmeans(results,args$clustercount)

copycluster <- function(which.cluster) {
  files<-photos[clusters$cluster==which.cluster]
  todir <- paste0(args$prefix,which.cluster)
  if (!isTRUE(file.info(todir)$isdir)) dir.create(todir, recursive=TRUE)

  file.copy(from = files,  to = paste0(todir,'/',files ))
}

cat(paste0("Copying photos into their groups..."))
a_ply(1:args$clustercount,1,copycluster)

Os scripts instalam as bibliotecas necessárias. No Linux / Mac, ele usará todos os núcleos disponíveis (no Windows, infelizmente não pode).

Deve funcionar imediatamente em qualquer sistema, e no Linux / Mac, obrigado pelo shebang, ele pode ser executado diretamente do terminal.