After I trained a word detection model on edge impulse. I downloaded the model in h5 and convert it to a model.cc file which is similar to the example from arduino micro_speech model.cpp. After i replace the model data and run the program. I found that it cant run it successfully. Is it the right way to do it if I dun want to use the zip which generated by edge impulse? Thanks a lot
Well, by downloading the .h5 you will only obtain the trained model not the full pipeline (that includes the DSP pre-processing).
What I invite you to do, is to go to the deployment page and then select the C++ library, there you will get everything in the generated .zip file.
Please, have a look at this documentation page to learn more about using the generated cpp library:
Let me know if you have any questions.
On which target do you intend to deploy your custom ML model?
Thanks Louis. I tried to do this before. However, I would like to combine two models in a same program which is hard to do this if I use the zip file from edge impulse. Therefore, I would like to know that can I download the model data directly from edge impulse, then I can replace the model data to the example of micro_speech in arduino nano library example.