test.lua

require 'torch'
require 'nn'
require 'image'
require 'paths'
require 'lib/AdaptiveInstanceNormalization'
require 'lib/utils'

local cmd = torch.CmdLine()

-- Basic options
cmd:option('-style', '',
           'File path to the style image, or multiple style images separated by commas if you want to do style interpolation or spatial control')
cmd:option('-styleDir', '', 'Directory path to a batch of style images')
cmd:option('-content', '', 'File path to the content image')
cmd:option('-contentDir', '', 'Directory path to a batch of content images')
cmd:option('-vgg', 'models/vgg_normalised.t7', 'Path to the VGG network')
cmd:option('-decoder', 'models/decoder.t7', 'Path to the decoder')

-- Additional options
cmd:option('-contentSize', 512,
           'New (minimum) size for the content image, keeping the original size if set to 0')
cmd:option('-styleSize', 512,
           'New (minimum) size for the style image, keeping the original size if set to 0')
cmd:option('-crop', false, 'If true, center crop both content and style image before resizing')
cmd:option('-saveExt', 'jpg', 'The extension name of the output image')
cmd:option('-gpu', 0, 'Zero-indexed ID of the GPU to use; for CPU mode set -gpu = -1')
cmd:option('-outputDir', 'output', 'Directory to save the output image(s)')
cmd:option('-saveOriginal', false, 
            'If true, also save the original content and style images in the output directory')

-- Advanced options
cmd:option('-preserveColor', false, 'If true, preserve color of the content image')
cmd:option('-alpha', 1, 'The weight that controls the degree of stylization. Should be between 0 and 1')
cmd:option('-styleInterpWeights', '', 'The weight for blending the style of multiple style images')
cmd:option('-mask', '', 'Mask to apply spatial control, assume to be the path to a binary mask of the same size as content image')

opt = cmd:parse(arg)

print(opt)
if opt.gpu >= 0 then
    require 'cudnn'
    require 'cunn'
end

assert(opt.content ~= '' or opt.contentDir ~= '', 'Either --content or --contentDir should be given.')
assert(opt.style ~= '' or opt.styleDir ~= '', 'Either --style or --styleDir should be given.')
assert(opt.content == '' or opt.contentDir == '', '--content and --contentDir cannot both be given.')
assert(opt.style == '' or opt.styleDir == '', '--style and --styleDir cannot both be given.')
assert(paths.filep(opt.decoder), 'Decoder ' .. opt.decoder .. ' does not exist.')

vgg = torch.load(opt.vgg)
for i=53,32,-1 do
    vgg:remove(i)
end
local adain = nn.AdaptiveInstanceNormalization(vgg:get(#vgg-1).nOutputPlane)
decoder = torch.load(opt.decoder)

if opt.gpu >= 0 then
    cutorch.setDevice(opt.gpu+1)
    vgg = cudnn.convert(vgg, cudnn):cuda()
    -- vgg:cuda()
    adain:cuda()
    -- decoder = cudnn.convert(decoder, cudnn):cuda()
    decoder:cuda()
else
    vgg:float()
    adain:float()
    decoder:float()
end

local function styleTransfer(content, style)

    if opt.gpu >= 0 then
        content = content:cuda()
        style = style:cuda()
    else
        content = content:float()
        style = style:float()
    end

    styleFeature = vgg:forward(style):clone()
    contentFeature = vgg:forward(content):clone()

    if opt.mask ~= '' then -- spatial control
        assert(styleFeature:size(1) == 2) -- expect two style images
        local styleFeatureFG = styleFeature[1]
        local styleFeatureBG = styleFeature[2]
        local C, H, W = contentFeature:size(1), contentFeature:size(2), contentFeature:size(3)
        local maskResized = image.scale(mask, W, H, 'simple')
        local maskView = maskResized:view(-1)
        local fgmask = torch.LongTensor(torch.find(maskView, 1)) -- foreground indices 
        local bgmask = torch.LongTensor(torch.find(maskView, 0)) -- background indices
        
        local contentFeatureView = contentFeature:view(C, -1)
        local contentFeatureFG = contentFeatureView:index(2, fgmask):view(C, fgmask:nElement(), 1) -- C * #fg
        local contentFeatureBG = contentFeatureView:index(2, bgmask):view(C, bgmask:nElement(), 1) -- C * #bg

        targetFeatureFG = adain:forward({contentFeatureFG, styleFeatureFG}):clone():squeeze()
        targetFeatureBG = adain:forward({contentFeatureBG, styleFeatureBG}):squeeze()
        targetFeature = contentFeatureView:clone():zero() -- C * (H*W)
        targetFeature:indexCopy(2, fgmask ,targetFeatureFG)
        targetFeature:indexCopy(2, bgmask ,targetFeatureBG)
        targetFeature = targetFeature:viewAs(contentFeature)

    elseif opt.styleInterpWeights ~= '' then -- style interpolation
        assert(styleFeature:size(1) == #styleInterpWeights, '-styleInterpWeights and -style must have the same number of elements')
        targetFeature = contentFeature:clone():zero()
        for i=1,styleFeature:size(1) do
            targetFeature:add(styleInterpWeights[i], adain:forward({contentFeature, styleFeature[i]}))
        end
    else
        targetFeature = adain:forward({contentFeature, styleFeature})
    end

    targetFeature = targetFeature:squeeze()
    targetFeature = opt.alpha * targetFeature + (1 - opt.alpha) * contentFeature
    return decoder:forward(targetFeature) 
end

print('Creating save folder at ' .. opt.outputDir)
paths.mkdir(opt.outputDir)

if opt.mask ~= '' then
    mask = image.load(opt.mask, 1, 'float') -- binary mask
end

local contentPaths = {}
local stylePaths = {}

if opt.content ~= '' then -- use a single content image
    table.insert(contentPaths, opt.content)
else -- use a batch of content images
    assert(opt.contentDir ~= '', "Either opt.contentDir or opt.content should be non-empty!")
    contentPaths = extractImageNamesRecursive(opt.contentDir)
end

if opt.style ~= '' then 
    style_image_list = opt.style:split(',')
    if #style_image_list == 1 then
        style_image_list = style_image_list[1]
    end
    table.insert(stylePaths, style_image_list)
else -- use a batch of style images
    assert(opt.styleDir ~= '', "Either opt.styleDir or opt.style should be non-empty!")
    stylePaths = extractImageNamesRecursive(opt.styleDir)
end

if opt.styleInterpWeights ~= '' then
    styleInterpWeights = opt.styleInterpWeights:split(',')
    local styleInterpWeightsSum = torch.Tensor(styleInterpWeights):sum()
    for i=1,#styleInterpWeights do -- normalize weights so they sum to 1
        styleInterpWeights[i] = styleInterpWeights[i] / styleInterpWeightsSum
    end
end

local numContent = #contentPaths
local numStyle = #stylePaths
print("# Content images: " .. numContent)
print("# Style images: " .. numStyle)


for i=1,numContent do
    local contentPath = contentPaths[i]
    local contentExt = paths.extname(contentPath)
    local contentImg = image.load(contentPath, 3, 'float')
    local contentName = paths.basename(contentPath, contentExt)
    local contentImg = sizePreprocess(contentImg, opt.crop, opt.contentSize)

    for j=1,numStyle do -- generate a transferred image for each (content, style) pair
        local stylePath = stylePaths[j]
        if type(stylePath) == "table" then -- style blending
            styleImg = {}
            styleName = ''
            for s=1,#stylePath do
                local style = image.load(stylePath[s], 3, 'float')
                styleExt = paths.extname(stylePath[s])
                styleName = styleName .. '_' .. paths.basename(stylePath[s], styleExt)
                style = sizePreprocess(style, opt.crop, opt.styleSize)
                if opt.preserveColor then
                    style = coral(style, contentImg)
                end
                style = style:add_dummy()
                table.insert(styleImg, style)
            end
            styleImg = torch.cat(styleImg, 1)
            styleName = styleName:sub(2)
        else
            styleExt = paths.extname(stylePath)
            styleImg = image.load(stylePath, 3, 'float')
            styleImg = sizePreprocess(styleImg, opt.crop, opt.styleSize)
            if opt.preserveColor then
                styleImg = coral(styleImg, contentImg)
            end
            styleName = paths.basename(stylePath, styleExt)
        end

        local output = styleTransfer(contentImg, styleImg)

        local savePath = paths.concat(opt.outputDir, contentName .. '_stylized_' .. styleName .. '.' .. opt.saveExt)
        print('Output image saved at: ' .. savePath)
        image.save(savePath, output)

        if opt.saveOriginal then
            -- also save the original images
            image.save(paths.concat(opt.outputDir, contentName .. '.' .. contentExt), contentImg)
            image.save(paths.concat(opt.outputDir, styleName .. '.' .. styleExt), styleImg)
        end
    end
end