RaspberryP​iへの転移学習された​モデルの展開方法につ​いて/How do I deploy a transfer-learned model to a Raspberry Pi?

resnet50の畳み込みネットワークを自身のデータで転移学習させ、そのmatファイルをresNet.matとして保存した。

転移学習の際には、下のコードでresnet50の最後の3つの層に手を加えてトレーニングさせた。

I used the pre-trained model, Resnet50, and trained my own images on it to create a network to differentiate between two categories.

When I did the transfer learning, I changed the last 3 layers of the Resnet50 model with the code listed below.

lgraph = removeLayers(lgraph, {'ClassificationLayer_fc1000','fc1000_softmax','fc1000'});
numClass = 2;

newLayers =[
    fullyConnectedLayer(numClass,'Name','fc','WeightLearnRateFactor',10,'BiasLearnRateFactor',10)
    softmaxLayer('Name','softmax')
    classificationLayer('Name','classoutput')];

I downloaded the MATLAB support package for the Raspberry pi and put the deep learning linux image on it.

I also confirmed that I could connect to the Raspberry Pi from my computer.

However, when I try to deploy the MAT file that I created with the code below, I come up with an error message.

The error message reads 'Invalid MAT file. MAT file should contain a single instance of either a SeriesNetwork, DAGNetwork, yolov2ObjectDetector or ssdObjectDetector object.'

Maybe I'm not understanding this error code properly, but I thought if I used transfer learning on one of these 4 networks, that the MAT file would be valid. That is why I used Resnet50 which is a DAGNetwork.

deployの関数でRasPiに展開しようとすると、

mynet = coder.loadDeepLearningNetwork('resNet.mat');

のコードでエラーとなった。そのエラーメッセージを下の画像に示す。

matファイルが無効だとエラーになったが、DAGNetworkを転移学習させているため、resNet.matもDAGNetwork

になり、エラーの条件を満たしていると考えていますが、

matファイルが無効な理由、RasPiに展開できない理由はなぜなのでしょうか？

I would appreciate any help.