Great work from Yi-An Chen with an extension of the work done on the FREAK detector to be even closer to the way the human eye works. Amazingly implementing a digital version of the physical layout of the human eye turns out to be an incredibly efficient arrangement for scale, rotation and colour independent feature detection. This continues to improve on the state of the art for computer vision and really shows how well tuned our human visual system is.

http://worldcomp-proceedings.com/proc/p2016/IPC3500.pdf

We should have some demos of this running inside the Umajin machine vision stack soon.