About challenge metric, SSIM

It says on the website(http://fastmri.org/submission_guidelines) that "The metric is calculated using the measure.compare_ssim function from the skimage python module with default parameters.’

The default parameters of compare_ssim are below.
compare_ssim(X, Y, win_size=None, gradient=False, data_range=None, multichannel=False, gaussian_weights=False, full=False, **kwargs)

But, on the github(https://github.com/facebookresearch/fastMRI/blob/master/common/evaluate.py), it says that
def ssim(gt, pred):
return compare_ssim(
gt.transpose(1, 2, 0), pred.transpose(1, 2, 0), multichannel=True, data_range=gt.max()

That is, “multichannel=True, data_range=gt.max()” is different from default parameters.
Which is correct?


Good catch. The SSIM values are computed using the code that you linked to. We will update the submission guidelines to reflect that.