Size of test dataset: 35

Percentage of successful automatic results: 90%

Percentage of less successful automatic results: 10%

Comparison with Grabler09: Since this manipulation does not include any selections of labeled regions, the Grabler09 result is identical to the Average result.