List of a predefined deep learning layers

I come across the question when I was learning the YOLO V2 example. In the example, it is recommended to use activation_40_relu as featureLayer in the resnet50. It seems that the layer "activation_40_relu" is a predefined layer in Matlab. My questions are:

1. Where can I find the detailed description for this layer?
2. I assume there are many such predefined layer to choose. Where can I find the list of such predefined layers, if I want to use another layer instead of "activation_40_relu"?

Answer 1

[Cris LaPierre]

I don't know if there is a detailed explanation somewhere of the ResNet-50 network.This is a pretrained network that can be added to MATLAB through the Add-ons Explorer, but you may need to look to the original creators of the network for specific details.

There are 177 total layers in the ResNet-50 model, corresponding to a 50 layer residual network (before adding the YOLO v2 layers).

After installing the Deep Learning Toolbox Model for ResNet-50 Network, open the Deep Network Designer. In the startup screen, select ResNet-50.

The model will open. You will see a visual representation of the model.

You can then zoom in to inspect the layers. Here is activation_40_relu. Any options for that layer will appear on the right.

You may find the Getting Started with Deep Network Designer page helpful.

[Cris LaPierre]

Found 2 links you might be interested in

• List of Deep Learning Layers
• Specify Layers of Convolutional Neural Network