It's been a while since my last post, I was in a bunch of compromises and I decided to travel last two weekends to loose, my routine was mind sucking and talk to some great people I haven't met for some months and I must say that this quick chunking recharged my batteries.
I don't know why but I decided to take a break on the 3D Modeling but that I mentioned on the last posts, I think I crossed against a small project that I thought I could conclude on a single week but this project is taking more than that.
I'm using C# and AForge (computer vision library for C#) in order to do body tracking and collision with virtual objects in real-time video. I got stuck on some Asynchronous problems and Marshalling but I could move on the problem I'm getting now is a very understandable one: performance.
Imagine a 30 fps video file coming to you and you want to apply all filters on it. You have 33 ms per function to do everything before a new frame comes in (1 second = 1000 ms; time per function = 1000 ms / 30 = approx. 33 ms). I'm getting 222 ms per function, it won't be an easy job to do that in C# (I know C/C++ would be faster, but I would've to waste many time on UI and things like that and AForge provides lots of great functions and support that made me use C#).
I'm posting the current result of my processing, these images are completely test and won't be used on the final project, so just pay attention on the idea itself which is: remove the background and provide corner detection for future usage on collision detection.


And that's it.
Next days I intend to finish this and resume the 3D Modeling book plus study more to the Master appliance test on Electrical Engineering.
See you,
Caio.
Next days I intend to finish this and resume the 3D Modeling book plus study more to the Master appliance test on Electrical Engineering.
See you,
Caio.