Looking for help with image- processing.

I am working a program that is supposed to correctly identify and label shapes in an image (square, triangle, circle, rectangle). I am working with a dataset of 9 images:

https://drive.matlab.com/sharing/4138e732-4f9b-4324-8fe7-75d836e3ff24

The dataset includes 2D images of shapes and 3D images of shapes.

The outline for the program is as follows:

• Reading images and pre-processing to workable level.
• Identify non accident properties and segment regions of concavity.
• Develop equations to define geometric primitives.
• Matching components to objective representation and object identification.

I have tried many techniques already in order to achieve this program including using equations to identify the shapes but i have not had much luck. On my lates version of my code I am using blob analysis which has enabled me to identify the 2D shapes with about 80% accuracy. I am also having a lot of trouble with the 3D shapes as they are all different colour and they create shadows and some are touching eachother.

I am fairly new to MatLab and coding in general. I am not sure if I am going the right way about it with the code I am currenty using. If anyone could look at my code and the dataset and maybe give me some example code to achieve the goal of the program, that would be a great help.

Unfortunately, I was not able to upload my code as a file or the dataset (MatLab told me i have exceeded my daily upload of 10 files even though have never uploaded a file before). I have included the link to the dataset above and the code is below.

% Read in the image file
originalImage = imread('1.png');

% Extract the red, green, and blue color channels
redChannel = originalImage(:, :, 1);
greenChannel = originalImage(:, :, 2);
blueChannel = originalImage(:, :, 3);

% Threshold values to isolate objects effectively
levelr = 0.20;
levelg = 0.20;
levelb = 0.31;

% Binarise each layer and add them together
redBinarize = imbinarize (redChannel, levelr);
greenBinarize = imbinarize (greenChannel, levelg);
blueBinarize = imbinarize (blueChannel, levelb);
rgbSum = (redBinarize&greenBinarize&blueBinarize);

% Compliment the image and fill in holes (Black as background, white as
% foreground)
icomp = imcomplement (rgbSum);

% Edge detection using Canny method
edgeImage = edge(icomp, 'Canny');

% Morphological operations to enhance edge detection
se = strel('square', 3);
edgeImage = imdilate(edgeImage, se);
edgeImage = imfill(edgeImage, 'holes');

% Blob analysis to find regions and their properties
blobAnalysis = regionprops(edgeImage, 'all');

% Loop through each region and label shapes
figure;
imshow(icomp);
hold on;
for k = 1:length(blobAnalysis)
    % Get region properties
    thisBlob = blobAnalysis(k);
    perimeter = thisBlob.Perimeter;
    area = thisBlob.Area;
    circularity = 4 * pi * area / perimeter^2;
    centroid = thisBlob.Centroid;
    
    % Determine shape based on circularity
    if circularity >= 0.9
        % Blob is circular
        shape = 'circle';
    elseif circularity < 0.9 && circularity >= 0.7
        % Blob is square or rectangle
        ratio = thisBlob.MinorAxisLength / thisBlob.MajorAxisLength;
        if ratio >= 0.9 && ratio <= 1.1
            % Blob is square
            shape = 'square';
        else
            % Blob is rectangle
            shape = 'rectangle';
        end
    elseif circularity < 0.7 && circularity >= 0.5
        % Blob is triangle
        shape = 'triangle';
    else
        % Blob is irregular
        shape = 'irregular';
    end
    
    % Label shape on image
    text(centroid(1), centroid(2), shape, 'Color', 'r', 'FontWeight', 'bold');
end

[Rena Berman]

(Answers dev) @zero, I have attached the images so you can reference them and run your code in the editor.