Flame detection in video using Artificial Neural Networks

Hai I am Implement Flame detection in video. For this first I apply back differential method on each frame. After this step I convert frame from RGB to HSV. Now after this I convert HSV to seprate channel H,S and V. Now I want to this H,S and V to give Artificial Neural Network toolBox to detect my flame pixel. Now I am very stuk in this point:how to prepare my data for Neural networks.What function is best as activation function.and what number of layers will be best for this? Kindly help me?

Answer 1

[Greg Heath]

When in doubt, use defaults.

 help patternnet
 doc patternnet

Hope this helps.

Greg