def test_AllImagesFromSampeSampleAreGrouped(self):
itemlist = []
for sampleID in ["001", "015", "230"]:
itemlist.append([
imageNamed(
int(sampleID) - 1, "%s_%d.png" % (sampleID, (i) + 1))
for i in range(27)
])
items = self.joinLists(itemlist)
numpy.random.shuffle(items)
target = FileGroupingDatasetSplitter(9)
result = target.split(items, [.6, .2, .2])
totalAfterSplit = self.joinLists(result)
self.assertEqual(sorted(totalAfterSplit), sorted(items))
for sampleID in ["001", "015", "230"]:
self.assertSamplesAreInOnlyOneSet(
["%s_%d.png" % (sampleID, i) for i in range(1, 10)], result)
self.assertSamplesAreInOnlyOneSet(
["%s_%d.png" % (sampleID, i) for i in range(10, 19)], result)
self.assertSamplesAreInOnlyOneSet(
["%s_%d.png" % (sampleID, i) for i in range(19, 28)], result)
def test_SplitsImages(self):
items = [
imageNamed(0, "001_01.bmp"),
imageNamed(0, "001_03.bmp"),
imageNamed(0, "001_02.bmp")
]
target = FileGroupingDatasetSplitter(1)
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 FileGroupingSplit(numFilePerImage, shuffle=True):
return FileGroupingDatasetSplitter(numFilePerImage, shuffle)
def CreateWithFileGroupingSplitter(imageSource, numFilePerImage, preprocessor = None):
return DatasetCreator(imageSource = imageSource,
datasetSplitter = FileGroupingDatasetSplitter(numFilePerImage),
preprocessor = preprocessor)
