I don't think that File::Find is going to let you throw away directory subtrees that you don't like. You could put code into your wanted() method that will ignore files with a path of the form that you describe, but you're going (to the best of my understanding) end up having File::Find waste time by walking through whole subtrees that you'd rather ignore. You might try using the following idiom in the place of File::Find to accomplish what you want...

use strict;
use Cwd;

my $cwd = Cwd::getcwd();
my $directory = shift(@ARGV) || $cwd;
$directory = $cwd . '/' . $directory unless $directory =~ /^\//;
my @queue = ($directory);

while (@queue) {
    my $node = shift(@queue);

    if (-d $node) {
        opendir(DIR, $node);
        push(@queue, map { $node . '/' . $_ } 
                     grep { $_ ne '.' and $_ ne '..' and $_  !~ /^_/ }
                     readdir(DIR));
        closedir(DIR);
    }
    else {
        do_your_stuff($node);
    }
}
[download]

This will do a depth first search on either the current working directory, or the directory that you specify on the command line, and it will ignore all subtrees of directories beginning with _. I hope this helps, and I hope it doesn't turn out to be a horribly convoluted way of doing it if there is an easier way with File::Find.

Update: It was come to my attention that exploitation of the $File::Find::prune variable is a much more expeditious way of accomplishing a pruning. I confess! I confess! Now stop minus one-ing me, I admit the error of my ways.