Sign In
Create Account
Follow Us on Twitter
Subscribe on Youtube
As a hobby I have been writing a program that will scan images and determine if the pictures is pornographic or not. What I use so far is the color of the image hue but the problem is there are over 1,000 colors of skin tone. I'd like to create an agorithm that can detect if there is a face in the picture as well as detect skin color. If more than 70% of the picture is skin color and there is a face I'd like to consider it porn.
Now, for my problem. How do I detect a face? I don't know where to begin. Can someone point me in the correct direction?
4 replies to this topic
Back to top|
|
|
#2
Posted 24 October 2006 - 08:55 AM
WingedPanther
-
- Moderators
-
- 16,831 posts
A spammer's worst nightmare
This is one of the areas of AI research. Things that pop out in my mind: eyes and mouth. Two areas with distinctly different colors present.
#3
Guest_Jordan_*
Posted 29 October 2006 - 11:25 AM
Guest_Jordan_*
Guest_Jordan_*
-
- Guests
This is a hard thing to do - I'd suggest reading as many reports on algorithms for this type of thing that you can find. A quick search on google didn't yield to many results.
#4
Posted 30 October 2006 - 09:49 AM
Paradine
-
- Members
-
- 48 posts
Learning Programmer
WingedPanther said:
This is one of the areas of AI research. Things that pop out in my mind: eyes and mouth. Two areas with distinctly different colors present.
Good point about the eyes and ears. I'd imagine there is an algorithm waiting to be made that could determine if they are position right to be a face as well. The color is a good idea as well.
#5
Posted 31 October 2006 - 10:20 PM
TcM
-
- Members
-
- 11,147 posts
Writes binary right handed and hex left handed
Well about the face one I saw a program that found a face ( and made a border with it ) when you loaded a picture! So it can be made but sorry i dont remember the program!
