question about evaluators.py

[Kevin35Day]

hi!

in the open-reid/reid/evaluators.py

def pairwise_distance(features, query=None, gallery=None, metric=None):
if query is None and gallery is None:
n = len(features)
x = torch.cat(list(features.values()))
x = x.view(n, -1)
if metric is not None:
x = metric.transform(x)
dist = torch.pow(x, 2).sum(1) * 2
dist = dist.expand(n, n) - 2 * torch.mm(x, x.t())
return dist

the result seems not euclidean metric? in this 'pairwise_distance' function,the second part is computing euclidean metric,but the first part looks not right? why they are different?

thanks!

[Cysu]

Say we have a matrix X with N samples. The first part actually computes the pairwise distances inside X, resulting in a NxN distance matrix.

Say we have another matrix Y with M samples. The second part actually computes the pairwise distance across X and Y, resulting in a NxM matrix.