Hello Morteza,
I understand that you want to know how to work with the ‘gTruth’ type object obtained from the Image Labeler App. The ground truth object from the Image labeler App contains the following properties:
- DataSource - this contains the source of the image i.e the image file path.
- LabelDefinitions - this contains the label definitions (name, description etc)
- LabelData - this contains the coordinates of bounding boxes in each image.
You can use this ‘gTruth’ to create training data for your network.
- For creating datastores or table from your ground truth for object detection, use the ‘objectDetectorTrainingData’ function.
- For creating datastores for semantic segmentation, you can use the ‘pixelLabelTrainingData’ function.
Alternatively, you can directly export your data in the form of ‘table’ from the Image Labeler app instead of ‘gTruth’ and this can be used as training data for networks.
You can write the data about bounding boxes coordinates(position) into a ‘.txt’ file using the ‘fopen’, ‘fprintf’ and ‘fclose’ functions.
Refer to the following documentation for better understanding:
- ‘gTruth’ type object- https://www.mathworks.com/help/vision/ref/groundtruth.html
- objectDetectorTrainingData: https://www.mathworks.com/help/vision/ref/objectdetectortrainingdata.html
- pixelLabelTrainingData: https://www.mathworks.com/help/vision/ref/pixellabeltrainingdata.html
- 'Image Labeler’App: https://www.mathworks.com/help/vision/ug/get-started-with-the-image-labeler.html
- File input/output functions: https://www.mathworks.com/help/matlab/low-level-file-i-o.html?s_tid=CRUX_lftnav
Hope this helps!
