Some question about PSNR metric

[BigDong]

Codes:

mmeval/mmeval/metrics/psnr.py

Lines 71 to 100 in e2f6dbf

     def add(self, predictions: Sequence[np.ndarray], groundtruths: Sequence[np.ndarray], channel_order: Optional[str] = None) -> None:  # type: ignore # yapf: disable # noqa: E501 
         """Add PSNR score of batch to ``self._results`` 
  
         Args: 
             predictions (Sequence[np.ndarray]): Predictions of the model. 
             groundtruths (Sequence[np.ndarray]): The ground truth images. 
             channel_order (Optional[str]): The channel order of the input 
                 samples. If not passed, will set as :attr:`self.channel_order`. 
                 Defaults to None. 
         """ 
  
         if channel_order is None: 
             channel_order = self.channel_order 
  
         for prediction, groundtruth in zip(predictions, groundtruths): 
             assert groundtruth.shape == prediction.shape, ( 
                 f'Image shapes are different: \ 
                     {groundtruth.shape}, {prediction.shape}.') 
             groundtruth = reorder_and_crop( 
                 groundtruth, 
                 crop_border=self.crop_border, 
                 input_order=self.input_order, 
                 convert_to=self.convert_to, 
                 channel_order=self.channel_order) 
             prediction = reorder_and_crop( 
                 prediction, 
                 crop_border=self.crop_border, 
                 input_order=self.input_order, 
                 convert_to=self.convert_to, 
                 channel_order=self.channel_order) 

Seems channel_order will not be used if channel_order is not None

way to fix:

maybe change self.channel_order -> channel_order is OK?

[LeoXing1996]

@BIGWangYuDong, thanks for your reporting! I will check all metrics related to low-level vision and fix this problem as soon as possible!

[Zaida Zhou]

closed by #89