Using machine learning techniques, computer scientists and materials scientists in Saarbrücken have now developed a method that is much more accurate and objective than conventional quality control procedures. Their results have just been published in Scientific Reports.