how to implement CBAM with a pretarined network in matlab. i have seen tutorials doing this on other platforms but not on matlab.

%this is the code i found on GitHub but couldn't implement it

lgraph = layerGraph();
tempLayers = imageInputLayer([227 227 3],"Name","imageinput");
lgraph = addLayers(lgraph,tempLayers);

tempLayers = maxPooling2dLayer([5 5],"Name","maxpool_1","Padding","same");
lgraph = addLayers(lgraph,tempLayers);

tempLayers = averagePooling2dLayer([5 5],"Name","avgpool2d_1","Padding","same");
lgraph = addLayers(lgraph,tempLayers);

tempLayers = [
    multiplicationLayer(2,"Name","multiplication_1")
    fullyConnectedLayer(10,"Name","fc_1")
    fullyConnectedLayer(10,"Name","fc_2")
    fullyConnectedLayer(10,"Name","fc_3")
    leakyReluLayer(0.01,"Name","leakyrelu_1")];
lgraph = addLayers(lgraph,tempLayers);

tempLayers = averagePooling2dLayer([5 5],"Name","avgpool2d_2","Padding","same");
lgraph = addLayers(lgraph,tempLayers);

tempLayers = maxPooling2dLayer([5 5],"Name","maxpool_2","Padding","same");
lgraph = addLayers(lgraph,tempLayers);

tempLayers = [
    multiplicationLayer(2,"Name","multiplication_2")
    leakyReluLayer(0.01,"Name","leakyrelu_2")
    sigmoidLayer("Name","sigmoid")
    yolov2OutputLayer([16 16;32 32],"Name","yolov2-out")];
lgraph = addLayers(lgraph,tempLayers);

% 清理辅助变量
clear tempLayers;

%链接所有层
lgraph = connectLayers(lgraph,"imageinput","maxpool_1");
lgraph = connectLayers(lgraph,"imageinput","avgpool2d_1");
lgraph = connectLayers(lgraph,"maxpool_1","multiplication_1/in2");
lgraph = connectLayers(lgraph,"avgpool2d_1","multiplication_1/in1");
lgraph = connectLayers(lgraph,"leakyrelu_1","avgpool2d_2");
lgraph = connectLayers(lgraph,"leakyrelu_1","maxpool_2");
lgraph = connectLayers(lgraph,"avgpool2d_2","multiplication_2/in1");
lgraph = connectLayers(lgraph,"maxpool_2","multiplication_2/in2");

lgraph

lgraph = 
  LayerGraph with properties:

     InputNames: {'imageinput'}
    OutputNames: {'yolov2-out'}
         Layers: [14x1 nnet.cnn.layer.Layer]
    Connections: [15x2 table]

[Tian,HCong]

Hello, have you resolved your problem now? I am also very interested in this problem.

[Laraib]

No i was not able to resolve it. Though my fellow implemented it but using tensorflow and python. In matlab, I couldn't do it.

[Shubho]

This worked for me. You an try running the same in the MATLAB 2024b release.