def test_ImageScaler(tmpdir): input_height = 32 input_width = 32 channels = 3 image = np.ones([channels, input_height, input_width]).astype(np.float32) scalar = 1.5 bias = [10, 20, 30] model = C.image_scaler(image, scalar, bias) verify_no_input(model, tmpdir, 'ImageScaler_0') x = C.input_variable(np.shape(image)) model = C.image_scaler(x, scalar, bias) verify_one_input(model, image, tmpdir, 'ImageScaler_1')
def test_ImageScaler(tmpdir, dtype): with C.default_options(dtype = dtype): input_height = 32 input_width = 32 channels = 3 image = np.ones([channels, input_height, input_width]).astype(dtype) scalar = 1.5 bias = [10, 20, 30] model = C.image_scaler(image, scalar, bias); verify_no_input(model, tmpdir, 'ImageScaler_0') x = C.input_variable(np.shape(image)) model = C.image_scaler(x, scalar, bias); verify_one_input(model, image, tmpdir, 'ImageScaler_1')