/diy/ - Do It Yourself

>>164886
I think this is more of a /sci/ question unless you are asking how to build the laser scanner yourself.

take a picture?

i see a lot of problems with a potential laser pointed at the sky, particularly when there may be issues with planes in your area

OP
i figure the laser it cant be damaging to anyone or thing because Stanford used a laser scanner in their self driving car

>>164886

If you can't easily fix some sort of sensor to the subject, like every bird or every fish, then you need to develop some sort of computer vision scanning software to keep track of the subjects and do data analysis on the subjects.

No doubt this was done by thousands of computer science graduate students somewhere. If you really want to do this, then I suggest you google the fuck out of computer vision research and see what you can find.

>>164952
this. tbh it's way too much for a single person to do in a reasonable amount of time. if you're really digging on machine vision, buy a kinect and do some simpler tracking stuff, then apply to grad school and do it for real.

I work with some very expensive totaling stations that can do what you ask, however there are problems encountered with non-stationary objects. Also, data-sets can be extraordinarily large (think 2 gigs per "picture").

Have you ever tried doing it yourself?

Like, seriously bro, diy; mkay? This shit is so easy my cat can do it...and he got hit by a semi just yesterday!

>>164923
>only self driving car uses laser
>laser surgery doesn't do anything else besides point to the area
>car manufactures and a bunch of tards know how to use a laser fully and its effects.

the only lidar that's gonna slew fast enough to catch meaningful frames of flocks of birds would be something like a velodyne, and the angular error you're gonna get from that is arguably going to be too large to accurately deduce velocity vectors unless you're particuarly close with it or they aren't too close together

>>164886
>> laser scanning of flocks of starlings
no, but back in the day I think some researchers did video analysis of flock behavior, as in with their own eyes.

Results freaked the researchers out too, once they learned that there wasn't a leader, they started coming up with all sorts of crazy ideas, like that all the animals in the flock were linked together with some sort of psychic connection. But that was before we really understood what was going on....


Laser scanning sounds like a bad idea for this sort of thing. Because you have a bunch of things moving fast, with somethings behind other things. Most laser scanners, can't scan very fast, due to the scanning mechanism. Sure you could use flash Lidar units or depth cameras, but those are fucking expensive. Plus, you'd need several of these expensive sensors, so the birds wouldn't be hidden in the shadows of other birds.

What you want to do is motion capture. In motion capture you take at least 3 cameras mounted perpendicularly to each other, like top, front, side, and take video. By looking at the 3 pictures of say a bright LED light or retroreflective strip moving through space, you can figure out where the strip is in space. With more cameras you can get higher definition of where the strip is.

According to one of my professors, it's possible to make your own motion capture system with off-the-shelf hardware, but one of the problems with off the shelf hardware is getting all cameras synchronized.

The other problem you face is that your setup will probably need to be huge, especially if you want to analyze birds. As in, you'd need a stadium huge.