Hacker News new | past | comments | ask | show | jobs | submit login
The Ancient Secrets of Computer Vision: The Evolution of Human Vision (fritz.ai)
83 points by addlesee 18 days ago | hide | past | web | favorite | 9 comments



In our research, we show that the fovea observed many times in biology may be an emergent property that emerged from an attention system:

https://bair.berkeley.edu/blog/2017/11/09/learn-to-attend-fo...


"Interestingly, Octopuses have very similar eyes but do not have a blind-spot. This is because our optic nerve comes out of the retina into the eye and then back out, whereas optic nerves in an octopus come out in the opposite direction."

This "design flaw", i.e. placing wiring (nerves) in front of detectors (retina) is interesting and has been discussed at length (e.g. https://www.quora.com/Is-Dawkins-right-about-the-human-eye-b...), because it is point of contention for people arguing for Intelligent Design. Cephalod eyes evolved differently so don't have this problem (impossible not to feel a bit jealousy for this, like the randomized gradient decent optimization designed by your colleague converging on a much better solution than yours :-)


I don’t understand how it is a point of contention for people arguing for intelligent design. Wouldn’t it fall in the same category as questions like “if God is so good why are there wars and poverty”, where the answer is “to your feeble human mind it would seem like a design flaw, but God knows everything and everything He does is perfect but because you’re not God you can’t understand his brilliance”?


I think so. Another way to say this is "if you think you observed a flaw in humans, it's by definition not a flaw cause god doesn't do those". Like heart attacks, childhood brain cancer, blind-spot in our eyes.


Reminds me of that octo-human hybrid imprisoned in a robot that looks like a giant walking smartphone from "Twenty-First Century in Twenty-One Words or Less":

https://1.bp.blogspot.com/-hqhjRSSvrKk/WgulKWrSrpI/AAAAAAAAW...


>Cephalod eyes evolved differently so don't have this problem

What exactly is the ‘problem’? How is your vision adversly effected by this layout? Or, are you just arguing about aesthetics?


This is a lovely Quanta article about how mammals had to survive by being nocturnal, and how the eyes once adjusted for dark condidtions could not evolve back to being used in daylight:

https://www.quantamagazine.org/hyperuniformity-found-in-bird...

Now what would be darn nifty would be replacing the somewhat deficient mammalian optical system with something like the avian system. Would that it were so simple to splice a few genes over...

I loved this article on so many levels, although I need a bit of a run up to get the hang of hyperuniformity (which I think is the same as random matricies)


Good read, but AFAICT researches still don’t fully understand how birds fly, yet planes fly at higher altitudes and at faster speeds. Computer vision is probably not going to mimic human eye that closely


This is sort of true. Birds are significantly more energy efficient than planes for some kinds of flight. Birds are also able to perform maneuvers that planes cannot such as landing on a branch. For transportation we just don't care about these features.

I agree that CV does not need to mimick everything about human vision but we still have a long way to go before having vision as robust and efficient as humans. Moreover there are benefits in interpretability and explainability to mimicking human vision more closely. If CV perceived the world similarly it can fail in more similar and predictable ways.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: