Why use a negative mean? What is the below function?

[DachunKai]

        self.negmean = self._move_to_device(torch.Tensor([x * -1 for x in mean]).view(3, 1, 1), self.device)
        revNormalize = transforms.Normalize(mean=negmean, std=std)
        self.TP = transforms.Compose([revNormalize])