Single recall score for all images

[colaber2]

Hello,
I was wondering how to get a single recall score for all the images?
Because when I use MeanAveragePrecision class with extended_summary=True it gives multiple values, like this:

'recall': tensor([[[[0.2922, 0.8148, 0.8148], [0.1226, 0.6415, 0.6415], [0.3920, 0.9440, 0.9440], [0.7500, 1.0000, 1.0000]]]], dtype=torch.float64)}

Thank you in advance.