NN Training: Add LFW validation for #100.

training/main.lua

@@ -28,6 +28,7 @@ torch.manualSeed(opt.manualSeed)
 paths.dofile('data.lua')
 paths.dofile('model.lua')
 paths.dofile('train.lua')
+paths.dofile('test.lua')
 paths.dofile('util.lua')
 
 if opt.peoplePerBatch > nClasses then
@@ -41,5 +42,6 @@ epoch = opt.epochNumber
 
 for _=1,opt.nEpochs do
    train()
+   test()
    epoch = epoch + 1
 end

training/opts.lua

@@ -40,6 +40,7 @@ function M.parse(arg)
    -- GPU memory usage depends on peoplePerBatch and imagesPerPerson.
    cmd:option('-peoplePerBatch', 15, 'Number of people to sample in each mini-batch.')
    cmd:option('-imagesPerPerson', 20, 'Number of images to sample per person in each mini-batch.')
+   cmd:option('-testBatchSize', 800, 'Batch size for testing.')
 
    ---------- Model options ----------------------------------
    cmd:option('-retrain', 'none', 'provide path to model to retrain with')

training/test.lua

@@ -0,0 +1,64 @@
+-- Copyright 2016 Carnegie Mellon University
+--
+-- Licensed under the Apache License, Version 2.0 (the "License");
+-- you may not use this file except in compliance with the License.
+-- You may obtain a copy of the License at
+--
+--     http://www.apache.org/licenses/LICENSE-2.0
+--
+-- Unless required by applicable law or agreed to in writing, software
+-- distributed under the License is distributed on an "AS IS" BASIS,
+-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-- See the License for the specific language governing permissions and
+-- limitations under the License.
+
+require 'io'
+require 'string'
+require 'sys'
+
+local lfwDir = '../data/lfw/aligned'
+
+local batchRepresentBin = "../batch-represent/main.lua"
+local lfwEvalBin = "../evaluation/lfw.py"
+
+local testLogger = optim.Logger(paths.concat(opt.save, 'test.log'))
+
+local function getLfwAcc(fName)
+    local f = io.open(fName, 'r')
+    io.input(f)
+    local lastLine = nil
+    while true do
+    local line = io.read("*line")
+    if line == nil then break end
+    lastLine = line
+    end
+    io.close()
+    return tonumber(string.sub(lastLine, 6, 11))
+end
+
+function test()
+   if opt.cuda then
+      model = model:float()
+   end
+   local latestModelFile = paths.concat(opt.save, 'model_' .. epoch .. '.t7')
+   local outDir = paths.concat(opt.save, 'lfw-' .. epoch)
+   print(latestModelFile)
+   print(outDir)
+   local cmd = batchRepresentBin
+   if opt.cuda then
+      cmd = cmd .. ' -cuda '
+   end
+   cmd = cmd .. ' -batchSize ' .. opt.testBatchSize ..
+      ' -model ' .. latestModelFile ..
+      ' -data ' .. lfwDir ..
+      ' -outDir ' .. outDir
+   os.execute(cmd)
+
+   cmd = lfwEvalBin .. ' Epoch' .. epoch .. ' ' .. outDir
+   os.execute(cmd)
+
+   lfwAcc = getLfwAcc(paths.concat(outDir, "accuracies.txt"))
+   testLogger:add{
+      ['lfwAcc'] = lfwAcc
+   }
+end

training/train.lua

@@ -102,10 +102,9 @@ function train()
       cutorch.synchronize()
    end
 
-   -- set the dropouts to training mode
    model:training()
    if opt.cuda then
-      model:cuda() -- get it back on the right GPUs.
+      model:cuda()
    end
 
    local tm = torch.Timer()