I organized the cross-frame data and found that many pairs of images are not similar.
I see that the similarity threshold used in the code is 0.6 in here.
So, Is there any recommended threshold to maintain similarity while trying not to reduce the amount of data, thanks!
I organized the cross-frame data and found that many pairs of images are not similar.
I see that the similarity threshold used in the code is 0.6 in here.
So, Is there any recommended threshold to maintain similarity while trying not to reduce the amount of data, thanks!