Assuming the hex string isrepresents UTF-8, you could do this:

use warnings;
use strict;
use open qw/:std :utf8/;
use HTML::Entities qw/decode_entities/;

my $x = '리작';
my $xd = decode_entities($x);

my $y = '\\xeb\\xa6\\xac\\xec\\x9e\\x91';
(my $yd = $y) =~ s/\\x([0-9a-f]{2})/chr hex $1/eig;
utf8::decode($yd);

print "$x =>$xd<=\n";
print "$y =>$yd<=\n";

if ( $xd eq $yd ) { print "Match!\n" }
else { die "Mismatch!\n" }
[download]

Output:

&#47532;&#51089; =>리작<=
\xeb\xa6\xac\xec\x9e\x91 =>리작<=
Match!

Update: I realized it's unclear to me whether your original string was player=리작 or player=&#47532;&#51089; - the latter happens on PerlMonks when you put Unicode characters into <code> tags. You'd have to use <pre> or <tt> tags instead, but then characters that have special meanings (either as HTML or by the PerlMonks engine) have to be escaped, such as <=&gt;, &=&amp;, [=&#91;, etc. If your original string was player=리작, then you don't need the decode_entities step - but it is necessary for the input string in $cfg{'default.player'} to have been properly decoded first though (you can check this e.g. with Devel::Peek, see also).