I’m currently deep diving into some new systems to integrate a machine learning process into ETL.
Writing JAF to keep up on skills. (Just another framework)
bad bad legacy code
I can honestly say, in all the companies I have ever worked for, there has only been one company that actually had stellar code! I can’t publicly say where, but it is my current employer.
Well, “bad bad legacy code” vs. “stellar code”… Ha! Nice mix…
I am currently studying Neural Networks and Deep-Learning. Seems a lot of these are pushing Python.
Never programmed Python before, so am looking more into the C versions and even PHP versions, too.
Anyone have experience in this area?
Currently installing LAMP on a RaspberryPi (3B+)… and installing certificates…etc… to be run on an IronMan HUD project… with overlay graphics/animations … and real-time external circuit (repulsors/arc reactor) updates (data) displayed on the HUD screen…
Here is a video of the first PoC (proof of concept)
Live feed and overlay tests:
No live feed going on in this one, but external data parsing/testing…
(you guys make me feel bad about myself reading the above!) LOL
Cool, what are you going to use as the image screen?
In the actual helmet?
I am going to be using a disassembled VuFine display… (little, hi-res, lcd display with a ‘aspheric concave fresnel lenses’, with the correct diopter rating)… has a direct (mini) HMDI input, which makes things much easier!
I designed some custom PCB’s as well for the repulsors, arc reactor…etc… so they are communicate (wire-lessly) with the HUD circuit (Arduino & RPi)
I have not tried to port things to an RPi Zero… (not sure if it has enough power?)… but I like the smaller footprint.
I really enjoy trying to bring together my ‘2’ worlds…
code and electronics (home automation, props…etc)
Wow, very fun stuff! I have a Pi3B+, but, have not decided what to do with it as yet…
Right now, it is got the Pi-Hole software on it. That works 100% great! Removes all ad feeds to all the
systems on the network. And, cuts down on internet bloat to speed up page loads. Don’t have it running
right now as I like to see the ad’s to see what is being pushed out to the world…
But, love that stuff, Whispers…
A web app for project/tender calculations with a lot of integrations to systems more or less (mostly less) accepting integrations.
A web app with some 3d js canvas where you can assemble semi complex solutions for water distribution. Never worked with 3d js so no idea how to load 3d models, move them around the canvas, snap together etc. Guess it will be fun.
I worked with an engineer that used one and have had it on my amazon list ever since!
They are neat… and SUPER small.
I’ll try and take a video this weekend of the project being displayed through it…
- still need to work out some sort of ‘bracket’ to secure it to the helmet.
One obstacle remains for where/how to mount the web cam?
In theory… there will already be electronics in the (outer) eye sockets in the helmet… (because they lit up/blink…etc)… so trying to use those to mount the camera in might be difficult?
I was thinking perhaps in the mouth? (but might really throw off the balance of watching video that is centered from your chest/chin?)
I picked up my VuFine for $90 bucks…
I have NOT committed to using one for each eye yet however…
The only thing about the VuFine is it goes over the glasses and is bulky a little depending on the glasses.
I was looking at the Sony glasses which I think are nicer, till I saw the price of about $900… Yikes!
You got yours for $90. Nice deal!
I dont have those issues.
My VuFine is:
1.) will be taken apart (no more housing), just pcb and display
2.) not used with glasses. (going inside of an Iron Man helmet)
Yep, you mentioned that. Nice! Post a pix of it when you get it set up. Curious to see it.
Interested in how it will work, too. So, keep us up to date on this thread…
I would keep it to a single eye, since the effect is to transpose the screen in a eye trick. Using two would actually limit vision.
Sounds like a very cool project and I would be interested in pics later as well!
There is really -no- vision through the helmet.
The eye holes are blocked with ‘led’ circuitry…
So the only ‘real’ vision is through the live cam feed… that will be displayed in the one eye now.
Still… for costs alone… I’m going to start with the ‘1’ eye only
If I get any motivation…Im actually going to try and implement some facial/object recognition (not any true AI stuff)… and see if that can add a little bit of ‘fun’ to the theme. (basically just shooting for when a person/face is recognized, just have some sort of techy 'box’folow the face around…etc)
If anyone every wants to get into Arduino stuff… let me know! Happy to share or help!
Whispers, I had downloaded some code that did facial recognition. It was free, I think on Github.
I never worked with it. Or, you could use Facebook’s type of code. I had found tons of free open source
code for facial recognition. Like OpenFace on Github: OpenFace So, you should be able to get that code. Depends a lot on what language you need it to be written in. Sounds like a fun project. Do a Google search on “open source facial recognition code” and you will find tons of apps for this part! Keep us up to date on this one! Thanks!
I wish I was open to use whatever software I wanted…
The problem (outside of knowledge alone… LOL) is…
I do not know of any platform that will:
1.) Do/provide a LIVE feed
2.) Allow overlays/animation
3.) Allows external communication for real-time display updates on the HUD display.
I’ll out myself here… hahaha…
but before I do… disclaimer.
*I know this is largely not supported.
*I know it has security issues.
But I am using Flash (Adobe) for the current video feed and overlay, animations.
Supports external webcam feeds
Allows me to do whatever animation or overlays I want (visuals, menus…etc)
I have installed LAMP… so I am using a mix of PHP and Flash (ActionScript) socket connections to communicate back and forth, to communicate between the ‘outside data’ and the RPi/Flash platform.
This being an off-line (so to speak) prop/project… the security stuff if not a huge issue/concern
** Same with my BarBot project! (I’ll post on that soon too!)
** Off to read up on OpenFace!
Thanks for the link
I just saw: Python
So already looks like it could be a replacement if I cant get the current Flash to ‘play ball’ lol
A quick 30 second check with my best friend Google found this example of HTML5 displaying a live webcam on a webpage: Live Cam Example Not sure if this is anything that can help, but, another way you could use a canvas and paint this page onto the canvas. Doing that would let you control various parts of the canvas as you deemed important. Like the face-scanner code displays or a RSS feed or other “live” feeds you might want on the screen…
Just an idea…