Use "evaluateD​etectionPr​ecision" function for multiclass objects

I have image processing data stored in a sturct format. I have also gnerated ground truth data using the image labler application. I am using struct2table function in order to convert the data to a table format. I am only converting the results that contain the bounding boxes, lablels and scores for each image. The issue I get is that each the fuction "evaluateDetectionPrecision" reaches data that contains labels for the detected part it returns an error

"Error using vision.internal.detector.evaluationInputValidation>checkDetectionResultsTable (line 80)

The label value in row 9 of the detection results table is undefined. It must be one of the following categorical

values: {Healthy_Insulator,Broken_Insulator,Dirty_Insulator}."

I am quite sure that the labels are in categroical format as it clearly shows. I am providng the mat files that contain the script " Ap_precs_test1.m" Ground_truth and results_test21.mat. I would highly appreciate help.