How can I visualize solov2 models like Alexnet,Resnet50,etc., in deep learning toobox

I want to see what kind of convolutional layers/pooling layers, etc., are used in this model, and how are these layers connected, in the author's original paper, and my knowledge level is limited to understand them all

Answer 1

[Malay Agarwal]

Hi @Yan,

MATLAB has a lot of these models built-in. Refer to the following resource for a list of the models available in MATLAB and how you can visualize them: https://www.mathworks.com/help/deeplearning/ug/pretrained-convolutional-neural-networks.html

Hope this helps!

[Yan]

Dear @Malay Agarwal

Thank you for your attention to this problem, but unfortunately, I did not find the structure of the solov2 model, only the way to use it.

[Peter Man]

@Yan, you can create an object detector in MATLAB like this:

detector = solov2("resnet50-coco");

However, as far as I can see there's no way to actually peek into the internal networks used in the object detector. But...if you look at the code for solov2 (by calling "edit solov2"), you will be able to see all the networks used there - see the internal properties such as FeatureNet, NeckNet, MaskFeatNet etc. Loading those networks into Deep Network Designer app might get you what you want.

[Yan]

Thanks!