def runtrain(args):
#config = LSTMIMG_LapConfig(load=True)
config = LSTMIMG_GPUConfig(load=True, args=args)
trainReader = LSTMIMGProcessor(config.trainAnnotFile,
config.rawQnTrain,
config.trainImgFile,
config,
is_training=True)
valReader = LSTMIMGProcessor(config.valAnnotFile,
config.rawQnValTestFile,
config.valImgFile,
config,
is_training=False)
#dumReader = DummyReader(config)
model = LSTMIMGmodel(config)
model.construct()
model.train(trainReader, valReader, config.logFile)
#model.train(dumReader, dumReader)
model.destruct()
return config
def validateInternalTestSet(args, restoreModel=None, restoreModelPath=None):
from vqaTools.vqaInternal import VQA
from vqaTools.vqaEval import VQAEval
#config = LSTMIMG_LapConfig(load=True, args)
config = LSTMIMG_GPUConfig(load=True, args=args)
if restoreModel is None:
restoreModel = config.restoreModel
restoreModelPath = config.restoreModelPath
print('Running Validation Test on Model')
valTestReader = LSTMIMGProcessor(config.testAnnotFile,
config.rawQnValTestFile,
config.testImgFile,
config,
is_training=False)
model = LSTMIMGmodel(config)
model.loadTrainedModel(restoreModel, restoreModelPath)
predFile = '{}PredsLSTMEverything.csv'.format(restoreModelPath)
results, strictAcc = model.runPredict(valTestReader, predFile)
model.destruct()
valTestReader.destruct()
print('predictions made')
vqa = VQA(config.testAnnotFileUnresolved, config.originalValQns)
vqaRes = vqa.loadRes(results, config.originalValQns)
vqaEval = VQAEval(vqa, vqaRes, n=2)
vqaEval.evaluate()
print('Writing to file..')
writeToFile(vqaEval, restoreModelPath, vqa, vqaRes, strictAcc)
def runMetricsForInternalTestSet(args, restoreModel, restoreModelPath):
#config = LSTMIMG_LapConfig(load=True, args)
config = LSTMIMG_GPUConfig(load=True, args=args)
print('Running Validation Test on Model')
valTestReader = LSTMIMGProcessor(config.testAnnotFile,
config.rawQnValTestFile,
config.testImgFile,
config,
is_training=False)
model = LSTMIMGmodel(config)
model.loadTrainedModel(restoreModel, restoreModelPath)
lab, pred, classToAnsMap = model.runEvaluationMetrics(valTestReader)
model.destruct()
valTestReader.destruct()
#run metrics & get stats
listOfStats = runMetrics(lab, pred, classToAnsMap, restoreModelPath)
#save to pickle
data = {}
data['labels'] = lab
data['preds'] = pred
data['classToAnsMap'] = classToAnsMap
dateID = restoreModelPath.split('/')[-2]
saveToPickle(data, 'labpreds{}.pkl'.format(dateID))
print('predictions made')
return listOfStats
def loadOfficialTest(args, restoreModel=None, restoreModelPath=None):
config = LSTMIMG_GPUConfig(load=True, args=args)
testReader = TestProcessor(qnFile=config.testOfficialDevQns,
imgFile=config.testOfficialImgFeatures,
config=config)
model = LSTMIMGmodel(config)
if restoreModel is None:
model.loadTrainedModel(config.restoreModel, config.restoreModelPath)
else:
model.loadTrainedModel(restoreModel, restoreModelPath)
model.runTest(testReader, config.testOfficialResultFile)
model.destruct()