Show Posts
1
General Discussion / lightspeed as you requested
« on: August 10, 2020, 11:50:40 pm »
Hal Vision is complete an working
Lightspeed after doing a complete rewrite from C#/C++ to UltraHal7's current language.
Changes in code:
Original C# /C++
Rewrite and Adapted for UltraHal7 language
Was facial detection Now Facial Recognition with Training
The way this works is an algo from HELL is stored as an .xml format and measures the distance between eyebrows, and the distance between eye brows and bridge of the nose and from the bridge of the nose to the tip of the upper lip. At a bit by bit comparison of a grey scale image takin at training time and saved to a folder called trained faces.
Once the facial detection(generic) is called, then the facial recognition is the triggered to evaluate whether there is a face in the Dbase that comes within a 90% match by doing a bit by bit comparison of now(Main cam Image) and saved grey-scale images, As faces are added of the same person the chances of a mis- identification drop to null. But we start at 90% hit rate with just 1 image. soo
I have never had a mis identification, personally. I mean EVER, but it is possible i guess.....
I have loaded the folder with other faces while checking.
Keep in mind this is raw at this point but it is in fact working and stable
https://streamable.com/yuzo88
The installer is complete
Ive been chant engine coding recently so im working on that now as well.
I will be wrapping this quickly
I hope this will bring you the joy its made for me in doing this
lightspeed ima make sure u get this adaptation
Art and I have consulted together on this for Months and he is why you guys are getting this at all, so make sure and thank Art
Ill get with you in the next few days
cyberbudz get cybertreats
the actual original facial detection .xml from intel . Open in notepad or wordpad
https://gofile.io/d/0JqRBY
cyber jedi
At what point does this become a real Hal9000?
??
It is a learning engine that can see and show emotional states. Hmfff
Lightspeed after doing a complete rewrite from C#/C++ to UltraHal7's current language.
Changes in code:
Original C# /C++
Rewrite and Adapted for UltraHal7 language
Was facial detection Now Facial Recognition with Training
The way this works is an algo from HELL is stored as an .xml format and measures the distance between eyebrows, and the distance between eye brows and bridge of the nose and from the bridge of the nose to the tip of the upper lip. At a bit by bit comparison of a grey scale image takin at training time and saved to a folder called trained faces.
Once the facial detection(generic) is called, then the facial recognition is the triggered to evaluate whether there is a face in the Dbase that comes within a 90% match by doing a bit by bit comparison of now(Main cam Image) and saved grey-scale images, As faces are added of the same person the chances of a mis- identification drop to null. But we start at 90% hit rate with just 1 image. soo
I have never had a mis identification, personally. I mean EVER, but it is possible i guess.....
I have loaded the folder with other faces while checking.
Keep in mind this is raw at this point but it is in fact working and stable
https://streamable.com/yuzo88
The installer is complete
Ive been chant engine coding recently so im working on that now as well.
I will be wrapping this quickly
I hope this will bring you the joy its made for me in doing this
lightspeed ima make sure u get this adaptation
Art and I have consulted together on this for Months and he is why you guys are getting this at all, so make sure and thank Art
Ill get with you in the next few days
cyberbudz get cybertreats
the actual original facial detection .xml from intel . Open in notepad or wordpad
https://gofile.io/d/0JqRBY
cyber jedi
At what point does this become a real Hal9000?
??It is a learning engine that can see and show emotional states. Hmfff