Open Leo-820 opened 3 years ago
M
and/or K
during training and testing. Specifically, the output will be similar as long as the receptive field is roughly the same. The network would still work even when the receptive field is significantly different, like twice larger or smaller. Of course there will be unexpected output when the difference is too extreme.
I wonder is the network able to use different
M
andK
value during training and testing? And if I understand correctly, the valueM
equal to the number of keypoint you got. So why USIP doesnot degenerate when detect only 4 keypoints (as experiments in Figure 4), since you mentioned that small value for M will increase the receptive field and cause the FPN to degenerate. Is it because you choose a proper value ofK
to avoid degeneration?