how to prepare the data set for boxLabelDatastore function using image labler APP

I have a data set prepared by image labler APP. The format of the bounding box is different to the example showing on the matlab page.

Object Detection Using YOLO v2 Deep Learning - MATLAB & Simulink - MathWorks Australia

I wrote some codes to convert them, but not very successful. There are either single or double quotation marks in my table and

The boxLabelDatastore report an error message as below:

Error using boxLabelDatastore>iAssertValidBBoxFormat

The size of bounding box data must be M-by-4, M-by-5, or M-by-9, where M is the number of boxes in each table element. The column in the training data table that contains the bounding boxes must be a cell array.

My questions are:

1) How could I convert them into a cell array without ' or ''

2) Is there a easy way to label image and produce the dataset that mataches the example data set?

Regards

data_snail = load('340Ann.mat');
imageFilename = data_snail.gTruth.DataSource;
% imageFilename = cell2table(imageFilename);
snaillist = data_snail.gTruth.LabelData.snail;

for i =1:length(snaillist)
    tmp=snaillist{i}; 
    rowStrings = arrayfun(@(row) strjoin(arrayfun(@(x) num2str(x), tmp(row, :), 'UniformOutput', false), ', '), 1:size(tmp, 1), 'UniformOutput', false);
    if size(tmp,1)>1
        resultString = strjoin(rowStrings, '; ');
    else
        resultString = rowStrings;
    end
    column(i,1) = string(imageFilename{i});
    resultString = strcat('[', resultString, ']');
    column(i,2) = string(resultString);
    resultString=[];
end

snailDataset = array2table(column, 'VariableNames', {'imageFilename', 'snail'});
snailDataset.snail=cellstr(snailDataset.snail);

Answer 1

[T.Nikhil kumar]

Hello Josh,

I understand that you want to create a “training Data” table from the “groundTruth” object generated by using the Image Labeler App to create a YOLOv2 object Detector.

This can be done in a simple way by using the “objectDetectorTrainingData” function that takes “groundTruth” type object and returns a “trainingDataTable” that can be used for implementation object detection using YOLOv2.

trainingDataTable = objectDetectorTrainingData(gTruth); 

You can refer to the following documentation to understand about the “objectDetectorTrainingData” function.

Create training data for an object detector - MATLAB objectDetectorTrainingData - MathWorks India

Hope this helps!