Hi, I was able to use and transfer learning from a pre-trained Convolutional Neural Network (CNN) following these two useful tutorials: The generated object is a 1x1 SeriesNetwork, which is able to classify my set of images with good accuracy. However, I would like to C Code this CNN to use outside Matlab. According to the Neural Network toolbox, I can use the command genFunction to generate a Matlab function that simulate the trained neural network and then use codegen to generate a C/C++ code. Probability Markov Chains Queues And Simulation Ebook Readers here. Fault Resilient Drivers For Longhorn Server Tips there. However, if I try to call genFunction I receive the following error: genFunction(convnet,'testConvNet') Undefined function 'genFunction' for input arguments of type 'SeriesNetwork'. My question is: Is there any other way to code this trained CNN to use it outside Matlab? *P.S. Dolby Digital Codec Not Found Premiere Elements 11 here. : 1- I also used bagOfFeatures function to generate features for my image classification problem. Although the accuracy is not as good as the one obtained with a CNN, I have the same problem, as the generated classifier is an imageCategoryClassifier object.

As far as I know, it is not supported by the C code tool. 2- There is a way to train a SVM classifier ( CompactClassificationECOC object), which is supported by the C coder.

But because in these tutorials it use the features calculated by the CNN convolutional layers, I still need the CNN convolutional layers to make the SVM work. I hope these details show that I did my homework before coming here. I have to admit my first contact with C coder tool was a week ago. Before I used to recode the application C after getting Matlab successful tests.