def test_from_normalised(self):
"Crops from normalised coordinates are as expected"
i = InselectImage(TESTDATA / 'test_segment.png')
h, w = i.array.shape[:2]
boxes = [Rect(0, 0, 1, 1), Rect(0, 0.2, 0.1, 0.8)]
self.assertEqual([Rect(0, 0, 459, 437), Rect(0, 87, 45, 349)],
list(i.from_normalised(boxes)))
def test_from_normalised(self):
"Crops from normalised coordinates are as expected"
i = InselectImage(TESTDATA / 'shapes.png')
h, w = i.array.shape[:2]
boxes = [Rect(0, 0, 1, 1), Rect(0, 0.2, 0.1, 0.8)]
self.assertEqual([Rect(0, 0, 459, 437), Rect(0, 87, 46, 350)],
list(i.from_normalised(boxes)))
def test_save_crop_overlapping(self):
"Save a crop that is partially overlapping the image"
i = InselectImage(TESTDATA / 'shapes.png')
temp = tempfile.mkdtemp()
try:
# A crop that is partially overlapping the image
p = Path(temp) / 'overlapping.png'
i.save_crops([Rect(-0.1, -0.1, 0.4, 0.3)], [p])
crop = InselectImage(p).array
# Crop should have this shape
self.assertEqual((131, 184, 3), crop.shape)
# Non-intersecting regions should be all zeroes
self.assertTrue(np.all(0 == crop[0:44, 0:46]))
self.assertTrue(np.all(0 == crop[0:44, ]))
self.assertTrue(np.all(0 == crop[:, 0:46]))
coords = list(i.from_normalised([Rect(-0.1, -0.1, 0.4, 0.3)]))
expected = i.array[0:87, 0:138, ]
self.assertTrue(np.all(expected == crop[44:, 46:, ]))
finally:
shutil.rmtree(temp)
def test_not_normalised(self):
i = InselectImage(TESTDATA / 'test_segment.png')
self.assertRaises(i.from_normalised([Rect(0, 0, 2, 2)]))
def test_not_normalised(self):
i = InselectImage(TESTDATA / 'test_segment.png')
self.assertRaises(i.from_normalised([Rect(0, 0, 2, 2)]))
