How DeepLab v3+ is connected with ResNet-18 in matlab?

In matlab how DeepLab v3+ is combined with ResNet-18? I mean how both architectures are combined to form deeplabv3plusLayers function? How ResNet-18 features are connected with DeepLab v3+ layers?

In this "https://www.mathworks.com/help/vision/ref/deeplabv3pluslayers.html" they explained how to create a DeepLab v3+ network based on ResNet-18. But how are they both combined to get features? (Which layers have to be connected from ResNet-18 to DeepLab v3+ to get this "deeplabv3plusLayers(imageSize,numClasses,'resnet18');" network)

Answer 1

[Srivardhan Gadila]

You can make use of analyzeNetwork function to see and analyze how resnet18 or other networks are being used in the deeplabv3plusLayers function.

resnet18Net = resnet18;
analyzeNetwork(resnet18Net)

imageSize = [224 224 3];
numClasses = 5;
network = 'resnet18';
deeplabv3Down16 = deeplabv3plusLayers(imageSize,numClasses,network,'DownsamplingFactor',16);
deeplabv3Down8 = deeplabv3plusLayers(imageSize,numClasses,network,'DownsamplingFactor',8);
analyzeNetwork(deeplabv3Down16)
analyzeNetwork(deeplabv3Down8)

[kumara dommeti]

Sir, I know how to analyze by using analyzeNetwork. But my question is how they are working to learn features when combined together? (I know how both networks are working when they are implemented seperately. But my question is when they are combined how they will learn features).