wdnmdqwer
wdnmdqwer
@0xfaded Thanks for your explanation and it's helpful. I have implemented your codes on my iPhone device and compared the fast algorithm to the OpenCV ones. But the result showed...
Thanks for your help. I compiled my project with release. It is surprising that the speed of your NEON code is faster than openCV version.
1. I highly recommend that non templated functions should be employed since in some cases image size cannot be known before compiling. Non templated functions could receive some parameters dynamically...
Hi @0xfaded . I have some new problems: 1. How do you check whether the consecutive 9 pixels around center meets the fast threshold requirement? 2. In your `fastDetect`, you...
1. Could I think that these piece of your codes only count the number of zeros from most significant bit? 2. I am trying to utilize NEON make orb extractor...
Yeah.
@dreamer1031 it is only for backend.