The code may have some bugs

[li624120638]

mmdet/models/data_preprocessors/data_preprocessor.py
DetDataPreprocessor._get_pad_shape():

elif isinstance(_batch_inputs, torch.Tensor):
    ....
    pad_h = int(np.ceil(_batch_inputs.shape[1] / self.pad_size_divisor)) * self.pad_size_divisor
    pad_w = int( np.ceil(_batch_inputs.shape[2] / self.pad_size_divisor)) * self.pad_size_divisor
    ....

might be

elif isinstance(_batch_inputs, torch.Tensor):
    ....
    pad_h = int(np.ceil(_batch_inputs.shape[2] / self.pad_size_divisor)) * self.pad_size_divisor
    pad_w = int( np.ceil(_batch_inputs.shape[3] / self.pad_size_divisor)) * self.pad_size_divisor
    ....

[hhaAndroid]

@li624120638 Yes, thank you very much for your feedback. May I ask if you can submit a pull request to fix it?