Error, Detector in YOLO V4

Hello everyone, can you help me? I am using the example "Object Detection Using YOLO v4 Deep Learning" in the document Matlab help, but with a change in the dataset (customs data). when I reach to detector appear the error.

this code :-

detector = yolov4ObjectDetector("csp-darknet53-coco",className1,className2,anchorBoxes,InputSize=inputSize);

this error appear:-

Error using yolov4ObjectDetector/parsePretrainedDetectorInputs
Expected a string scalar or character vector for the parameter name.

Error in yolov4ObjectDetector (line 218)
                params = yolov4ObjectDetector.parsePretrainedDetectorInputs(varargin{:});

Error in recognitionThree (line 113)
detector = yolov4ObjectDetector("csp-darknet53-coco",className1,className2,anchorBoxes,InputSize=inputSize);

Answer 1

[Walter Roberson]

detector = yolov4ObjectDetector("csp-darknet53-coco",className1,className2,anchorBoxes,InputSize=inputSize);

yolov4ObjectDetector

detector = yolov4ObjectDetector(baseNet,classes,aboxes,'DetectionNetworkSource',layer)

So in your call, the name parameter is being passed as "csp-darknet53-coco" . The classes parameter is being passed as className1 . The aboxes parameter is being passed as className2 . Then you get to the anchorBoxes parameter in your call, and since the positional parameter locations have been filled, the code has to match the contents of anchorBoxes to one of the option names. But your anchorBoxes variable is not a character vector or a string array, so you get an error message.

The classes has to be passed as a single parameter, such as {className1, className2}

[weam]

thank you so much for your explanation, and I am trying to solve the problem with your advices

[weam]

when i try to solve the problem by this way yolov4ObjectDetector , Create Custom YOLO v4 Object Detector example.

this code :-

classNames = {'Tool', 'Workpiece'};

basenet = resnet50;
analyzeNetwork(basenet)
basenet.Layers(1)
imageSize = basenet.Layers(1).InputSize;
layerName = basenet.Layers(1).Name;
newinputLayer = imageInputLayer(imageSize,'Normalization','none','Name',layerName);

lgraph = layerGraph(basenet);

lgraph = removeLayers(lgraph,'ClassificationLayer_fc1000');

lgraph = replaceLayer(lgraph,layerName,newinputLayer);

dlnet = dlnetwork(lgraph);

featureExtractionLayers = ["activation_22_relu","activation_40_relu"];

detector = yolov4ObjectDetector(dlnet,classNames,anchorBoxes,DetectionNetworkSource=featureExtractionLayers);

the error :

Error using yolov4ObjectDetector>iConfigureDetector
Number of anchor boxes must match the number of output layers.

Error in yolov4ObjectDetector/parseDetectorInputs (line 917)
                lgraph = iConfigureDetector(lgraph,numClasses,params.AnchorBoxes,params.DetectionNetworkSource);

Error in yolov4ObjectDetector (line 224)
                params = yolov4ObjectDetector.parseDetectorInputs(varargin{:});

Error in recognitionThree (line 141)
detector = yolov4ObjectDetector(dlnet,classNames,anchorBoxes,DetectionNetworkSource=featureExtractionLayers);

[Walter Roberson]

You do not show how you built anchorBoxes

[weam]

The anchorBoxes :-

anchors = anchors(idx,:);
anchorBoxes = {anchors(1:3,:)
    anchors(4:6,:)
    anchors(7:9,:)
    };

but when i change the anchorBoxes to this, the previos error is disappear, and new error appear:-

anchors = anchors(idx,:);
anchorBoxes = {anchors(1:5,:)
    anchors(6:9,:)
    };

this new error

Invalid transform function defined on datastore.

The cause of the error was:

Unable to use a value of type cell as an index.

Error in recognitionThree>augmentData (line 231)
    bboxes = A{ii,{1,1}};

Error in matlab.io.datastore.TransformedDatastore/applyTransforms (line 723)
                            data = ds.Transforms{ii}(data);

Error in matlab.io.datastore.TransformedDatastore/read (line 235)
            [data, info] = ds.applyTransforms(data, info);

Error in recognitionThree (line 155)
    data = read(augmentedTrainingData);
    
    

the augmentedTrainingData coded with supported function

    augmentedTrainingData = transform(trainingData,@augmentData);

augmentedData = cell(4,1);
for k = 1:4
    data = read(augmentedTrainingData);
    augmentedData{k} = insertShape(data{1},"rectangle",data{2}, data{4});
    reset(augmentedTrainingData);
end
figure
montage(augmentedData,BorderSize=10)


% Helper function for performing data augmentation.
function data = augmentData(A)
% Apply random horizontal flipping, and random X/Y scaling. Boxes that get
% scaled outside the bounds are clipped if the overlap is above 0.25. Also,
% jitter image color.

data = cell(size(A));
for ii = 1:size(A,1)
    I = A{ii,1};
    bboxes = A{ii,2};
    labels = A{ii,3};
    sz = size(I);

[Walter Roberson]

We need to see your complete function augmentData

[weam]

function data = augmentData(A);

data = cell(size(A)); for ii = 1:size(A,1) I = A{ii,1}; bboxes = A{ii,2}; labels = A{ii,3}; sz = size(I);

    if numel(sz) == 3 && sz(3) == 3
        I = jitterColorHSV(I,...
            contrast=0.0,...
            Hue=0.1,...
            Saturation=0.2,...
            Brightness=0.2);
    end

    % Randomly flip image.
    tform = randomAffine2d(XReflection=true,Scale=[1 1.1]);
    rout = affineOutputView(sz,tform,BoundsStyle="centerOutput");
    I = imwarp(I,tform,OutputView=rout);

    % Apply same transform to boxes.
    [bboxes,indices] = bboxwarp(bboxes,tform,rout,OverlapThreshold=0.25);
    labels = labels(indices);

    % Return original data only when all boxes are removed by warping.
    if isempty(indices)
        data(ii,:) = A(ii,:);
    else
        data(ii,:) = {I,bboxes,labels};
    end
end
end

[Walter Roberson]

the error message says

Error in recognitionThree>augmentData (line 231)
  bboxes = A{ii,{1,1}};

but that line does not exist in the augmentData function that you posted.

Answer 2

[Pritesh Shah]

Check input command of this function.