일반 이미지(height x widht x channel)에서 torch (channel x height x width)로 변환

xs = np.array([[[1, 2, 3], [4, 5, 6], [7, 8, 8]]])  # (1, 2, 3)
np.transpose(xs, (1, 2, 0))  # (2, 3, 1)