VolumetricUpSamplingNearest.lua

local VolumetricUpSamplingNearest, parent = torch.class('nn.VolumetricUpSamplingNearest', 'nn.Module')

--[[
Applies a 3D up-sampling over an input composed of several input planes.

The upsampling is done using the simple nearest neighbor technique.

The Y and X dimensions are assumed to be the last 2 tensor dimensions.  For
instance, if the tensor is 5D, then dim 4 is the y dimension and dim 5 is the x.

otime  = time*scale_factor
owidth  = width*scale_factor
oheight  = height*scale_factor
--]]

function VolumetricUpSamplingNearest:__init(scale_t, scale_xy)
   parent.__init(self)

   self.scale_factor_t = scale_t
   self.scale_factor_xy = scale_xy
   
   if self.scale_factor_t < 1 or self.scale_factor_xy < 1 then
     error('scale_factor must be greater than 1')
   end
   if math.floor(self.scale_factor_t) ~= self.scale_factor_t or math.floor(self.scale_factor_xy) ~= self.scale_factor_xy then
     error('scale_factor must be integer')
   end
   self.inputSize = torch.LongStorage(5)
   self.outputSize = torch.LongStorage(5)
   self.usage = nil
end

function VolumetricUpSamplingNearest:updateOutput(input)
   if input:dim() ~= 5 and input:dim() ~= 4 then
     error('VolumetricUpSamplingNearest only support 4D or 5D tensors')
   end
   -- Copy the input size
   local xdim = input:dim()
   local ydim = input:dim() - 1
   local tdim = input:dim() - 2
   for i = 1, input:dim() do
     self.inputSize[i] = input:size(i)
     self.outputSize[i] = input:size(i)
   end
   self.outputSize[tdim] = self.outputSize[tdim] * self.scale_factor_t
   self.outputSize[ydim] = self.outputSize[ydim] * self.scale_factor_xy
   self.outputSize[xdim] = self.outputSize[xdim] * self.scale_factor_xy
   -- Resize the output if needed
   if input:dim() == 4 then
     self.output:resize(self.outputSize[1], self.outputSize[2],
       self.outputSize[3],self.outputSize[4])
   else
     self.output:resize(self.outputSize)
   end
   input.nn.VolumetricUpSamplingNearest_updateOutput(self, input)
   return self.output
end

function VolumetricUpSamplingNearest:updateGradInput(input, gradOutput)
   self.gradInput:resizeAs(input)
   input.nn.VolumetricUpSamplingNearest_updateGradInput(self, input, gradOutput)
   return self.gradInput
end