Re: face detection
- From: azrael <jura.grozni@xxxxxxxxx>
- Date: Thu, 11 Oct 2007 09:38:47 -0000
A little mistake. Don't automaticaly decide how much neurons to use.
Take a program like Neuro solutions and give him a set of patterns and
he will give you a specific number of layers and neurons to use for
the best segmentation.
Esspecialy because usintg a backpropagation network, you don't need
more then one hidden layer. Look out for the Horner Theorem.
.
- References:
- Re: face detection
- From: siju
- Re: face detection
- Prev by Date: Re: face detection
- Next by Date: Is camera response needed for HDR based upon multiple exposures ?
- Previous by thread: Re: face detection
- Next by thread: Calling for suggestion on Super-Resolution
- Index(es):
Relevant Pages
|