MATLAB Answers


how to export ground truth labeler using image labeler app into table?

Asked by Shamitha Pillay on 11 Dec 2018
Latest activity Commented on by Kazuya Machida on 28 Nov 2019 at 3:12
When i run the example ('deeplearning_shared/DeepLearningFasterRCNNObjectDetectionExam*ple'), and load the 'fasterRCNNVehicleTrainingData.mat', it generates 'vehicleTrainingData with value 295*2 table.
When i label my own images using the ImagerLabeler app and export the file, it comes in groud Truth; gTruth with value 1*1 groundTruth...
How do I make it generate or load the data same as the example into table??.
What do i have to do in order to generate the same output value when loading data from image Labeler Ap


Sign in to comment.

1 Answer

Answer by Kazuya Machida on 8 Jan 2019

You can use "objectDetectorTrainingData" function to convert gTruth to table data after exporting gTruth from image labeler app. Could you refer to the document if you want to know how to use it (>>doc objectDetectorTrainingData).


If I use this method:
trainingData = objectDetectorTrainingData(gTruth,'SamplingFactor',2);
I always get this error:
"Error using objectDetectorTrainingData (line 156)
The input groundTruth object/s do not contain any valid object detector training data."
I have done the labeling with Matlabs Image Labeler and I exported the gTruth via "Export Labels" -> "to workspace".
What can I do?
Thank you very much!
Hi Vanessa,
What type of label did you use in your labeling? Rectangle or Pixel label?objectDetectorTrainingData is used for "object detection". So, this function requires rectangle labels. If there are no rectangle labels in your groundTruth object, you may see this types of errors.

Sign in to comment.