Consider using a hash of arrays (HoA) to track the partial paths (that includes all up to the file extension) and their associated types. For example, given your dataset, the hash would be:

'Dir1/001' => ['jpg'],
'Dir2/002' => ['jpg'],
'Dir2/001' => ['jpg', 'cr2']
[download]

This way you can track 'uniqueness' and the file types of the images:

use strict;
use warnings;
use File::Find::Rule;
use File::Basename;

my %hash;
my $dir = '.';

my @files = File::Find::Rule->file()->in($dir);

for my $file (@files) {
    if ( my ( $partialPath, $type ) = $file =~ /(.+)\.([^.]+)/ ) {
        push @{ $hash{$partialPath} }, $type;
    }

}

for my $partialPath ( keys %hash ) {
    #my @fileTypes = @{ $hash{$partialPath} };
    my $fileTypes = $hash{$partialPath}->[0] ? "@{ $hash{$partialPath}
+ }" : 'None';
    my $baseFilename = basename $partialPath;

    # make thumbnail
    # $partialPath contains all up to trailing dot and extension
    # @fileTypes contains the file type(s)
    # $baseFilename contains the base file name w/o the extension

    print "Partial Path: $partialPath\n";
    print "File Type(s): $fileTypes\nBasename: $baseFilename\n\n";
}
[download]

Hope this helps!