# Algorithm Behind Fast Feature detector

**URL:** <https://forum.opencv.org/t/algorithm-behind-fast-feature-detector/2911>\
**Category:** C++\
**Tags:** features2d\
**Created:** [April 22, 2021, 4:34am UTC](https://forum.opencv.org/t/algorithm-behind-fast-feature-detector/2911 "2021-04-22T04:34:37Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![sayan\_chaudhuri](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.opencv.org/sayan_chaudhuri/32/1891_2.png) [@sayan\_chaudhuri](https://forum.opencv.org/u/sayan_chaudhuri)\
**Post date:** [April 22, 2021, 4:34am UTC](https://forum.opencv.org/t/algorithm-behind-fast-feature-detector/2911/1 "2021-04-22T04:34:37Z")

</div>

I am trying to implement the FASt feature detector from scratch for my college assignments. I am following the paper of Dr. Rosten and have used the segment test for corner detection criterion. But, I find that FAST feature detector of OPENCV is finding corners better than my own implementation. So, I would like to know other than the segment test, is there anything else that is being checked in the FAST implementation of OPENCV?