problem while getting precision-recall curve

hi

i am ebaluation my faster rcnn detector to get curve and encounter with this problem :

Dot indexing is not supported for variables of this type.

Error in fasterRCNNtestdetect (line 57)

precision = metrics.ClassMetrics.Precision(classID);

MY CODE IS

clear
load('C:\Users\ZBook\OneDrive\Desktop\detector\detectorFasterRCNN.mat');
dataRoad=load("D:\this is.mat");
img=imageDatastore(dataRoad.gTruth.DataSource.Source);
labeldata=dataRoad.gTruth.LabelData;
blds=boxLabelDatastore(labeldata);
cds=combine(img,blds);
preview(cds)
tbl = countEachLabel(blds);
% Define the split ratios (e.g., 70% training, 15% validation, 15% testing)
trainRatio = 0.15;
valRatio = 0.15;
testRatio =  0.7;
% Count the total number of images
numImages = numel(img.Files );
% Calculate the number of images for each split
numTrain = round(trainRatio * numImages);
numVal = round(valRatio * numImages);
numTest = numImages - numTrain - numVal;
% Shuffle the datastore
cds = shuffle(cds);
inputSize = [224 224 3];
% Split the datastore into training, validation, and testing sets
trainingData = subset(cds, 1:numTrain);
validationData = subset(cds, numTrain+1:numTrain+numVal);
testData = subset(cds, numTrain+numVal+1:numTrain+numVal+numTest);
testdata = transform(testData,@(data)preprocessData(data,inputSize));
% Make predictions
% Make predictions
% Set MiniBatchSize to 1 for detection
detectionResults = detect(detector, testData, 'MinibatchSize', 1, 'Threshold', 0.1);
disp(detectionResults);
numClasses = 8;  % Replace with the actual number of classes in your dataset
% Loop through all class IDs
for classID = 1:numClasses
    % Make predictions
    % Set MiniBatchSize to 1 for detection
    
    % Evaluate the model for the current class
    metrics = evaluateDetectionPrecision(detectionResults, testData);
    
    % Extract precision and recall for the current class
    precision = metrics.ClassMetrics.Precision(classID);
    recall = metrics.ClassMetrics.Recall(classID);
    
    % Display or store the precision and recall for the current class
    fprintf('Class %d - Precision: %.2f, Recall: %.2f\n', classID, precision, recall);
    
    % Optionally, you can perform further analysis or visualization for each class
    % For example, you could plot precision-recall curves or confusion matrices for each class
    
    % Plot precision-recall curve for the current class
    figure
    plot(recall, precision)
    xlabel('Recall')
    ylabel('Precision')
    grid on
    title(sprintf('Precision-Recall Curve for Class %d', classID));
end
function data = preprocessData(data, targetSize)
% Resize image and bounding boxes to the targetSize.
sz = size(data{1}, [1, 2]);
scale = targetSize(1:2) ./ sz;
data{1} = imresize(data{1}, targetSize(1:2));
% Pass imageSize to helperSanitizeBoxes
imageSize = targetSize;  % You may adjust this based on your needs
data{2} = helperSanitizeBoxes(data{2}, imageSize);
% Resize boxes to new image size.
data{2} = bboxresize(data{2}, scale);
end

[Walter Roberson]

What shows up for

class(metrics)
class(metrics.ClassMetrics)

[ahmad]

I don't understand is this solution?

Answer 1

[Michael]

@ rainbow friends

Hello, I think you need to modify the code in your for loop where you're accessing the precision and recall values. Instead of using dot notation, you can use curly brackets {} to access the values. Here's the updated code:

matlab
    % Loop through all class IDs
    for classID = 1:numClasses
        % Make predictions
        % Set MiniBatchSize to 1 for detection
        
        % Evaluate the model for the current class
        metrics = evaluateDetectionPrecision(detectionResults, testData);
        
        % Extract precision and recall for the current class
        precision = metrics{1}.ClassMetrics.Precision(classID);
        recall = metrics{1}.ClassMetrics.Recall(classID);
        
        % Display or store the precision and recall for the current class
        fprintf('Class %d - Precision: %.2f, Recall: %.2f\n', classID, precision, recall);
        
        % Optionally, you can perform further analysis or visualization for each class
        % For example, you could plot precision-recall curves or confusion matrices for each class
        
        % Plot precision-recall curve for the current class
        figure
        plot(recall, precision)
        xlabel('Recall')
        ylabel('Precision')
        grid on
        title(sprintf('Precision-Recall Curve for Class %d', classID));
    end

By using curly brackets {}, you can access the first element of the metrics cell array and then access the precision and recall values using dot notation.

Answer 2

[Walter Roberson]

    metrics = evaluateDetectionPrecision(detectionResults, testData);

The function evaluateDetectionPrecision always returns a numeric value as the first output. The second and third outputs are potentially cell arrays.

Metrics are something returned from evaluateObjectDetection