Something really needs to be done about this embedded image Spam.  Honestly,
SpamBayes appears to be ineffective against all these images, and this kind
of e-mail represents almost 80% of what I'm receiving now.  The flood of
embedded-image based e-mail has skyrocketed.

Most of my image ads:
 - Drug ads
 - Weight Loss/Energy crap
 - Home Refinancing
 - Stock tips
 - Watch replicas

I'm sure OCR isn't the only way, but the words are there in plain view.  It
seems like the obvious way to resolve this.  It's SO frustrating, because
it's basically the exact same four or five images, just slightly modified
each time.  But they include all the words that would clobber this stuff if
it was text based.

SpamBayes has been such a great program for me and my colleges, family and
friends.  I can only hope that the project sees fit to resolve this soon.

FMJ

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On
Behalf Of Tony Meyer
Sent: Tuesday, October 04, 2005 10:24 PM
To: Vincent Zweije
Cc: [email protected]
Subject: Re: [Spambayes] Inspecting images (was: SpamBayes to HandleEmbedded
Images)

>> Other than a miraculous OCR feature showing up in SpamBayes soon, I'm 
>> out of ideas for a simple way of managing this type of mail on my 
>> home PC.
>
> Hmm... if anyone is going to work on this, do it the Bayesian way.
> Don't try to recognise text, just recognize patterns in the image.
> Just like spambayes doesn't understand language, just recognizes words 
> (and a little of their context).

FWIW, I suspect that this may be quite effective.  Techniques like this are
used in image processing already - for example, the Viola- Jones method of
detecting patterns (e.g. faces) in an image uses a cascade of haar-like
classifiers: essentially, a whole lot of tiny little classifiers that look
at tiny parts of the image (which are all automatically generated via
training, not specified as rules).

If only there was more time in the day ... :)

=Tony.Meyer

--
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.


_______________________________________________
[email protected]
http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html


_______________________________________________
[email protected]
http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html

Reply via email to