error while running graph=layerGraph(net)

Tan
2023 5 14
1 个回答
更新时间：2023 5 14
2 次查看（30 天）

0 个投票

graph = layerGraph(net);
Error using layerGraph
lgraph = layerGraph(net);
↑
Invalid argument at position 1. Input must be an array of layers, a SeriesNetwork, a DAGNetwork, or a dlnetwork.

2 个评论
显示 无 隐藏 无

Matt J
Matt J 2023-5-14
So what's the question? Presumably your 'net' input is none of the the legal input types listed in the error message.
Tan
Tan 2023-5-14
编辑：Walter Roberson 2023-5-14
net = googlenet('Weights','none');
%lys = net.Layers;
%lys(end-3:end)
numClasses = numel(categories(imdsTrain.Labels));
lgraph = layerGraph(net);
after that it show
Error using layerGraph
lgraph= layerGraph(net);
↑
Invalid argument at position 1. Input must be an array of layers, a SeriesNetwork, a DAGNetwork, or a dlnetwork.

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Matt J
Matt J 2023-5-14

0 个投票

Like I said, your 'net' is not one fo the types listed in the error message:
net = googlenet('Weights','none');
net =
LayerGraph with properties:
Layers: [144×1 nnet.cnn.layer.Layer]
Connections: [170×2 table]
InputNames: {'data'}
OutputNames: {'output'}

4 个评论
显示 2更早的评论 隐藏 2更早的评论

Tan
Tan 2023-5-14
so how should i solve this?
Walter Roberson
Walter Roberson 2023-5-14
By now, R2023a, googlenet() returns DAGNetwork and the R2023a layerGraph() is happy with it.
In R2020b, googlenet() is documented as returning DAGNetwork and the error message implies layerGraph in that version should accept it.
Matt J
Matt J 2023-5-14
编辑：Matt J 2023-5-14
Why is a solution even required? You can see googlenet() is giving you a LayerGraph direcly, without the need for further manipulation.

请先登录，再进行评论。

类别

在 帮助中心 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

产品

版本

R2020b

标签

提问:

Tan
Tan
2023-5-14

编辑：

Matt J
Matt J
2023-5-14

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by