Vision Thing

Enter Your Project for a chance to win an Oscilloscope Grand Prize Package for the Most Creative Vision Thing Project!

Back to The Project14 homepage

Project14 Home
Monthly Themes
Monthly Theme Poll

 

So its been a little bit since I got the beaglebone_ai so I thought I would finally do a post on what I have been working on.  So it all starts out of necessity when it comes to projects for me so with the bb:ai there was certainly a ton of things I could have done with it.

 

I first off started playing around again with it to get myself familiar with cloud9 again as it was a while I go I used it since I have been spending all my time on the raspberry pi.  I played around with the demo and a few other projects but I wanted to see how it would actually do compared to something like the pi and how well a nice program I found by Claude Pageau Sometimes we come across a project that well you just gotta respect so on to the project.

 

The first thing I did was got the beagle bone ai all up to date which didn't take too long to get it setup on wifi there is tons of tutorials on this and I think shabaz  did a great job on that in BeagleBone AI (BB-AI) - Getting Started

 

So the great thing about my project is getting it setup to where I could start messing around with it was very simple and wasn't that much that had to be done right out of the box a few tweaks here and there but nothing major to start getting data and results saved to my beaglebone ai.

 

Opening cloud9 I opened a terminal window and then typed:  cd [hit enter]

after that I ran the following code: curl -L https://raw.github.com/pageauc/speed-camera/master/speed-install.sh | bash

once all was said and one and installed it was really fun to get going I have to say this saved me a lot of time getting things going with the beaglebone ai.

then all I had to do was type: cd speed-camera/

after that:  ./menubox.sh

then I had to start the speedcam.py and webserver.py  you could alternatively run this from the terminal with:

python speedcam.py

and then in another terminal

python webserver.py

 

 

But I like the simplicity of the menubox that Claude made.

Not much left to go now that it has gone through the config and testing we can start to calibrate the speeds a bit better.

 

This is where I came across a few problems due to my location first of all I have 2 roads in front of my house one side road and then the highway.

So I could either get accurate readings at this point on one or the other but not both at the same time.  Which makes sense since the height and distance difference.

 

Time to launch the webserver main screen and see what we are greeted with.

Ok, so this will certainly get a revision in the future for me for sure.  but it does the trick. now on to seeing if we are actually recording some data lets head over to the images folder.

 

This is where I was able to get a acurate reading on the veichles going by my house the speed limit here is 50 kph so that looks pretty good as a reference to me on the frontage road here.

As you can see from the images I took from Claudes page there is  even more data we can go through when it comes to the webserver here.

 

But before I wrap up this project I wanted to show the result of the highway speeds when the frontage road is configured correctly.

 

So it should be about double the speed there based on the review of the photos that I have taken with the unit which is up in the 800 captures the last hour and half.  This guy was really hauling at about 105kph speed limit is 90kph there.

 

One thing I was really amazed at was its ability to even process the speeds based on the lights at night here is one of the images as you can see the speed is pretty dead on again. this is once again on the frontage road.

I am really impressed with the way that the beaglebone ai has handled this project there has been no hiccups and it processes each car quickly and moves onto the next in frame.

 

Now that I have it setup its most definitely going to get more and more attention in the coming months and will have to get some shots of myself driving by the unit at the speed limits to see if I can calibrate it even more and then redesign the web GUI as well a bit.

 

I would love to get this setup in a better location on my property probably just down in my carport so it cuts out the highway and focus's directly onto the frontage road.  There is a ton of other things yet to be played with in Claude's program but I love the ease of use and everything is laid out so things can be edited and adjusted and played with to ones heart content.

 

Make sure to check out Claudes project here: https://github.com/pageauc/speed-camera

 

So it was more responsive with the beaglebone ai but all the feature worked alot better right away on the raspberry pi when I ran it on there as well.

I hope you enjoyed the project and make sure to try it out yourself!