Code generation error (loadDeepL​earningNet​work)

I saw the example and followed.

https://kr.mathworks.com/help/supportpkg/raspberrypiio/ref/identify-objects-within-video-using-resNet-50-on-raspberry-pi-hardware.html?s_tid=srchtitle_raspberry%20resnet_1

However, I would like to replace my newly created resnet50 neural network, not the pre-trained resnet50 neural network provided by Matlab. I actually made it, and it works well on the Matlb. However, when deploying to Raspberry Pi, loadDeeplelearningnetwork function keeps failing. What's the cause?

Does it matter if my gpu is AMD? Matlab coder is NVIDIA compatible..

this is my function

this is deploy function

errors

[Ram Kokku]

@Chanhyeok, thank you for reaching out. I dont see any obvious issue with the code causing the error, but you may be able to work around the issue by passing a dummy character array as a second argument for coder.loadDeepLearningNetwork.

for example,

net = coder.loadDeepLearningNetwork('dataset800.mat', 'my_dummy_name');

Regarding your inquiry concerning AMD and NVIDIA GPUs, if your target is the Raspberry Pi or any other CPU-based platform, the GPU on the target platform or host is inconsequential.

[Chanhyeok]

Thank you for your kindness. I agree with your idea that a second argument causes error. I was wonder how to use second argument. Let me know additional explanation about second argument please, :) Is it okay to just use “my dummy name” or something has to be changed?

[Ram Kokku]

@Chanhyeok, The coders will use the second argument as the recommended name for the C++ class generated for the deep learning network. you can give a simple dummy character array, preferable a single word - for example, 'MyDLClass' or 'my_dummy_string'.

you can find more about this argument here : https://www.mathworks.com/help/gpucoder/ref/coder.loaddeeplearningnetwork.html#d124e9416

[Chanhyeok]

@Ram Kokku

I'm sorry that it doesn't work..

Answer 1

[Sayan Saha]

Hello @Chanhyeok, are you trying to codegen in a folder that has non-ASCII characters in it's path? If so, can you change to a directory that has only ASCII characters in it's full-path and see if that resolves the issue. Note that codegen is not currently supported for non-ASCII characters https://www.mathworks.com/help/coder/ug/code-generation-for-characters.html.

[wtg Wtg]

Hi Sayan Saha,

I run gpu checks on r2024a/b over Nvidia RTX4080 Laptop with all combinations of VS2017/2019/2022 + Cuda11.8-12.2/cuDNN8.7-9.2/TensorRT8.5.1.7-8.6.1.6, and always got the deepcodegen & deepcodeexec fails as below :

>>gpuEnvObj = coder.gpuEnvConfig('host');

gpuEnvObj.BasicCodegen = 1;

gpuEnvObj.BasicCodeexec = 1;

gpuEnvObj.DeepLibTarget = 'cudnn';

gpuEnvObj.DeepCodegen = 1;

gpuEnvObj.DeepCodeexec = 1;

results = coder.checkGpuInstall(gpuEnvObj)

Compatible GPU : PASSED

CUDA Environment : PASSED

Runtime : PASSED

cuFFT : PASSED

cuSOLVER : PASSED

cuBLAS : PASSED

cuDNN Environment : PASSED

Host Compiler : PASSED

Basic Code Generation : PASSED

Basic Code Execution : PASSED

Deep Learning (cuDNN) Code Generation: FAILED (GPU code generation failed with an error. View report for further information: View report)

results =

struct with fields:

gpu: 1

cuda: 1

cudnn: 1

tensorrt: 0

hostcompiler: 1

basiccodegen: 1

basiccodeexec: 1

deepcodegen: 0

tensorrtdatatype: 0

deepcodeexec: 0

The Error Report always shows as below :

Could you help ?