_single_size_image_list removing duplicates also removing the corresponding filename/label

[LIEeOoNn]

Is your feature request related to a problem?

When removing duplicates of an ImageList the ImageList length will be different from the filenames/ labels, thus causing problems when training a nn.

Desired solution

the remove_duplicate_images() meth should also have a parameter for the filenames/labels list[str] so both are updated correctly
solution which is also much faster than the old one

def remove_duplicate_images (self, filenames: list[str]) ->tuple[ImageList, list[str]]:
        import numpy
        image_list = self.to_images()
        image_list_without_dubs: list[Image] = []
        images = ImageList
        filenames_new: list[str] = []
        unique_byte = set()
        for i in range(len(image_list)):
            tensor = image_list[i]._image_tensor
            tensor_byte = tensor.numpy().tobytes()
            if tensor_byte not in unique_byte:
                unique_byte.add(tensor_byte)
                image_list_without_dubs.append(image_list[i])
                filenames_new.append(filenames[i])
        images = images.from_images(image_list_without_dubs)
        return images, filenames_new

Possible alternatives (optional)

No response

Screenshots (optional)

No response

Additional Context (optional)

No response