def test_dataset_length():
try:
train_image_loader = ImageLoader(
root_dir='proj2_code/proj2_unit_tests/small_data/',
split='train',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
test_image_loader = ImageLoader(
root_dir='proj2_code/proj2_unit_tests/small_data/',
split='test',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
except:
train_image_loader = ImageLoader(
root_dir='../proj2_code/proj2_unit_tests/small_data/',
split='train',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
test_image_loader = ImageLoader(
root_dir='../proj2_code/proj2_unit_tests/small_data/',
split='test',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
assert len(train_image_loader) == 30
assert len(test_image_loader) == 39
def test_load_img_from_path():
try:
test_image_loader = ImageLoader(
root_dir='proj2_code/proj2_unit_tests/small_data',
split='train',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
im_path = 'proj2_code/proj2_unit_tests/small_data/train/bedroom/image_0003.jpg'
except:
test_image_loader = ImageLoader(
root_dir='../proj2_code/proj2_unit_tests/small_data',
split='train',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
im_path = '../proj2_code/proj2_unit_tests/small_data/train/bedroom/image_0003.jpg'
im_np = np.asarray(test_image_loader.load_img_from_path(im_path))
try:
expected_data = np.loadtxt(
'proj2_code/proj2_unit_tests/test_data/sample_inp.txt')
except:
expected_data = np.loadtxt(
'../proj2_code/proj2_unit_tests/test_data/sample_inp.txt')
assert np.allclose(expected_data, im_np)
def test_unique_vals():
try:
train_image_loader = ImageLoader(
root_dir='proj2_code/proj2_unit_tests/small_data/',
split='train',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
except:
train_image_loader = ImageLoader(
root_dir='../proj2_code/proj2_unit_tests/small_data/',
split='train',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
item1 = train_image_loader.__getitem__(2)
item2 = train_image_loader.__getitem__(5)
assert not torch.allclose(item1[0], item2[0])
def test_class_values():
try:
test_image_loader = ImageLoader(
root_dir='proj2_code/proj2_unit_tests/small_data/',
split='test',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
except:
test_image_loader = ImageLoader(
root_dir='../proj2_code/proj2_unit_tests/small_data/',
split='test',
transform=get_fundamental_transforms(inp_size=(64, 64),
pixel_mean=np.array([0.01]),
pixel_std=np.array([1.001])))
class_labels = test_image_loader.class_dict
class_labels = {ele.lower(): class_labels[ele] for ele in class_labels}
expected_vals = {
'opencountry': 0,
'industrial': 1,
'office': 2,
'insidecity': 3,
'kitchen': 4,
'tallbuilding': 5,
'mountain': 6,
'forest': 7,
'store': 8,
'livingroom': 9,
'street': 10,
'bedroom': 11,
'coast': 12,
'suburb': 13,
'highway': 14
}
assert len(class_labels) == 15
assert set(class_labels.keys()) == set(expected_vals.keys())
assert set(class_labels.values()) == set(expected_vals.values())
def test_fundamental_transforms():
'''
Tests the transforms using output from disk
'''
transforms = get_fundamental_transforms(
inp_size=(100, 50), pixel_mean=[0.5], pixel_std=[0.3])
try:
inp_img = Image.fromarray(np.loadtxt(
'proj2_code/proj2_unit_tests/test_data/transform_inp.txt', dtype='uint8'))
output_img = transforms(inp_img)
expected_output = torch.load(
'proj2_code/proj2_unit_tests/test_data/transform_out.pt')
except:
inp_img = Image.fromarray(np.loadtxt(
'../proj2_code/proj2_unit_tests/test_data/transform_inp.txt', dtype='uint8'))
output_img = transforms(inp_img)
expected_output = torch.load(
'../proj2_code/proj2_unit_tests/test_data/transform_out.pt')
assert torch.allclose(expected_output, output_img)