target_types parameter of OxfordIIITPet() accepts the multiple same values

[hyperkai]

🐛 Describe the bug

target_types parameter of OxfordIIITPet() accepts the multiple same values as shown below. *It's not useful to have the multiple same elements:

from torchvision.datasets import OxfordIIITPet

my_data = OxfordIIITPet(
    root="data",
    target_types=["category", "binary-category", "segmentation", "category", "binary-category", "segmentation"]
)

trainval_cate_data[0]
# (<PIL.Image.Image image mode=RGB size=394x500>,
#  (0, 0, <PIL.PngImagePlugin.PngImageFile image mode=L size=394x500>,
#   0, 0, <PIL.PngImagePlugin.PngImageFile image mode=L size=394x500>))

Versions

import torchvision

torchvision.__version__ # '0.20.1'