Hello!
Thanks for your answer!
Perhaps try:
do ...
ciwd <- sequence classifiedImagesWithData
let allImages = (getImages ciwd)
...
as it seems like you're trying to gave getImages act on a value of
type [IO (ClassifiedImage, Image)]. Applying sequence to
classifiedImagesWithData will turn it into an IO [(ClassifiedImage,
Image)] by doing the natural thing, from which you can pull out the
list to apply getImages to.
I've tried that and wrote
<code-snippet>
do...
let classifiedImagesWithData = ((return trainingSet) >>= readClassifiedImages)
ciwd <- (sequence classifiedImagesWithData)
let allImages = (getImages ciwd)
</code-snippet>
Now, I get the error
<error>
Compiling ExperimentalYaleDb ( ExperimentalYaleFaceDb.hs, interpreted )
ExperimentalYaleFaceDb.hs:41:
Couldn't match `[]' against `IO'
Expected type: [[a]]
Inferred type: [IO (ClassifiedImage, Image)]
In the first argument of `sequence', namely
`classifiedImagesWithData'
In a 'do' expression: ciwd <- (sequence classifiedImagesWithData)
Failed, modules loaded: TestLik, Lik, HUnit, HUnitText, HUnitBase, HUnitLang.
</error>
at line
ciwd <- (sequence classifiedImagesWithData)
Are there any other options?
Thanks in advance
Dmitri Pissarenko
PS: Maybe the error is rooted in code parts other than those given here. In
the attachment there is the code of the main program
(ExperimentalYaleFaceDb.hs) and the function definitions (Lik.hs).
--
Dmitri Pissarenko
Software Engineer
http://dapissarenko.com
module ExperimentalYaleDb
where
import Lik
createTrainingSet = [(ClassifiedImage "../data-test/yalefaces-pgm/subject01.centerlight.pgm" "subject01"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject02.centerlight.pgm" "subject02"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject03.centerlight.pgm" "subject03"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject04.centerlight.pgm" "subject04"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject05.centerlight.pgm" "subject05"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject06.centerlight.pgm" "subject06"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject07.centerlight.pgm" "subject07"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject08.centerlight.pgm" "subject08"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject09.centerlight.pgm" "subject09"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject10.centerlight.pgm" "subject10")]
main = do
-- create a list of image to subject mapping
-- the training set is all centerlight images of subjects 1 to 10
--
-- Following images are not contained in the training set
--
-- a) images of subjects 11 to 15
-- b) images of subjects 1 to 10, which are not centerlight
let trainingSet = [(ClassifiedImage "../data-test/yalefaces-pgm/subject01.centerlight.pgm" "subject01"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject02.centerlight.pgm" "subject02"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject03.centerlight.pgm" "subject03"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject04.centerlight.pgm" "subject04"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject05.centerlight.pgm" "subject05"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject06.centerlight.pgm" "subject06"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject07.centerlight.pgm" "subject07"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject08.centerlight.pgm" "subject08"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject09.centerlight.pgm" "subject09"),
(ClassifiedImage "../data-test/yalefaces-pgm/subject10.centerlight.pgm" "subject10")]
trainingSet :: [ClassifiedImage]
-- read images from files
let classifiedImagesWithData = ((return trainingSet) >>= readClassifiedImages)
ciwd <- (sequence classifiedImagesWithData)
-- calculate average image
--let allImages = (getImages classifiedImagesWithData)
let allImages = (getImages ciwd)
-- calculate eigenfaces of the training set
-- calculate weights of the training set
-- now create a list of unknown images
-- these are
-- a) images of subjects 11 to 15
-- b) images of subjects 1 to 10, which are not centerlight
return 0module Lik
where
import IO
import Array
data ClassifiedImage = ClassifiedImage {imageFileName :: String, subjectID :: String}
deriving Show
--type ClassifiedImageWithData = IO (ClassifiedImage, IO Image)
type ClassifiedImageWithData = (ClassifiedImage, Image)
type Image = [Double]
--getImagesSelector :: Monad IO => IO (ClassifiedImage, Image) -> Image
getImagesSelector :: (ClassifiedImage, Image) -> Image
getImagesSelector classifiedImageWithData = snd classifiedImageWithData
getImages :: [(ClassifiedImage, Image)] -> [Image]
-- getImages classifiedImgs = (map getImagesSelector classifiedImgs)
getImages classifiedImgs = snd (unzip classifiedImgs)
readClassifiedImages :: [ClassifiedImage] -> [IO (ClassifiedImage, Image)]
readClassifiedImages classifiedImages = (map readClassifiedImagesSelector classifiedImages)
--readClassifiedImagesSelector :: ClassifiedImage -> IO (ClassifiedImage, Image)
readClassifiedImagesSelector classifiedImg = do
image <- (readImageAsDouble classifiedImg)
return (classifiedImg, image)
readImageAsDouble :: ClassifiedImage -> IO Image
readImageAsDouble classifiedImg = do
imageAsString <- readImage (imageFileName classifiedImg)
return (normalizeImg (convertToList imageAsString))
readImage :: String -> IO String
readImage filename = do
fileContent <- readFile filename
return fileContent
writeImage filename stuffToWrite = do
writeFile filename stuffToWrite
commentLine :: String -> Bool
commentLine text | ((head text) == '#') = True
| otherwise = False
noCommentLine :: String -> Bool
noCommentLine text = not (commentLine text)
removeComments :: String -> String
removeComments text = foldr (appendWithSeparator) "" linesWithoutComments
where linesWithoutComments = filter noCommentLine (lines text)
appendWithSeparator :: String -> String -> String
appendWithSeparator part1 part2 = whole
where whole = part1 ++ " " ++ part2
convertToArr :: forall a. (Read a) => String -> Array Int a
convertToArr fileContent = listArray (1, imgDatLength) imgDat
where imgDat = convertToList fileContent
imgDatLength = length imgDat
convertToList fileContent = drop 3 fileContentNumbers
where fileContentNoComments = removeComments fileContent
fileContentTokens = words fileContentNoComments
-- fileContentTokens contains token "P2", the magic number, which
-- we don't need
-- we "remove" it by taking the tail of fileContentTokens
fileContentNumbers = map read (tail fileContentTokens)
fileContentNumArr = listArray (1, length fileContentNumbers) fileContentNumbers
convertToFileContent :: Int -> Int -> Int -> [Int] -> String
convertToFileContent width height maxGray pixels = foldr (appendWithSeparator) [] (["P2", (show width), (show height), (show maxGray)] ++ pixelsAsStrings)
where pixelsAsStrings = map show pixels
normalizeImg :: [Int] -> Image
normalizeImg img = map normalizePixel imgAsDouble
where imgAsDouble = map fromIntegral img
normalizePixel :: Double -> Double
normalizePixel pixel = pixel/255
averageImage :: [Image] -> Image
averageImage images = map (/ numberOfImages) sum
where
sum = sumOfImages images
numberOfImages = fromIntegral(length images)
sumOfImages :: [Image] -> Image
sumOfImages images = foldr (add2Img) emptyImg images
where emptyImg = replicate (length (head images)) 0.0
add2Img :: Image -> Image -> Image
add2Img summand1 summand2 = sum
where sum = zipWith (+) summand1 summand2
denormalizeImg :: Image -> Image
denormalizeImg normalizedImage = map (*255) normalizedImage
doubleToInt :: Image -> [Int]
doubleToInt image = map (ceiling) image
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
