Closed ZeroSteven618 closed 4 years ago
Excuse me, I am sorry to bother you with the theoretical question.
During the LIFT pipeline, you got 4 patches based on VSFM, which is the patches samples label tool. The VSFM performs reconstruction with SIFT.
I an not sure about: Why the SIFT based training data can work so better than SIFT? Also better than the state-of-the-art methods.
Is it from the score map in KP part network?
Multiple things. But in a nutshell, since the network learns which points are best statistically from large data.
Excuse me, I am sorry to bother you with the theoretical question.
During the LIFT pipeline, you got 4 patches based on VSFM, which is the patches samples label tool. The VSFM performs reconstruction with SIFT.
I an not sure about: Why the SIFT based training data can work so better than SIFT? Also better than the state-of-the-art methods.
Is it from the score map in KP part network?