def test_AllImagesFromSampeSampleAreGrouped(self):
itemlist = []
samples = [("bark_00", 0), ("bark_01", 0), ("bark_02", 0),
("beans_01", 1), ("beans_02", 1), ("beans_03", 1)]
for sample in samples:
sampleID = sample[0]
sampleClass = sample[1]
itemlist.append([
imageNamed(sampleClass, "%s_%s.png" % (sampleID, imgType))
for imgType in ["o", "r", "rs", "s"]
])
items = self.joinLists(itemlist)
numpy.random.shuffle(items)
target = SampleGroupingDatasetSplitter(self.getNumberFunction)
result = target.split(items, [.3, .3, .4])
totalAfterSplit = self.joinLists(result)
self.assertEqual(sorted(totalAfterSplit), sorted(items))
for sample in samples:
sampleID = sample[0]
self.assertSamplesAreInOnlyOneSet([
"%s_%s.png" % (sampleID, imgType)
for imgType in ["o", "r", "rs", "s"]
], result)
def test_SplitsImages(self):
items = [
imageNamed(0, "bark_01_o.bmp"),
imageNamed(0, "bark_02_o.bmp"),
imageNamed(0, "bark_03_o.bmp")
]
target = SampleGroupingDatasetSplitter(self.getNumberFunction)
result = target.split(items, [.6, .2, .2])
self.assertEquals(sorted(self.joinLists(result)), sorted(items))
for d in result:
self.assertEquals(1, len(d))
def GroupingSplit(GetSampleNumberFunction = None):
if GetSampleNumberFunction is None:
GetSampleNumberFunction = GetSampleNumberFromFilename()
return SampleGroupingDatasetSplitter(GetSampleNumberFunction)
def CreateWithSampleGroupingSplitter(imageSource, preprocessor = None, GetSampleNumberFunction = None):
if GetSampleNumberFunction is None:
GetSampleNumberFunction = GetSampleNumberFromFilename()
return DatasetCreator(imageSource = imageSource,
datasetSplitter = SampleGroupingDatasetSplitter(GetSampleNumberFunction),
preprocessor = preprocessor)
