I am working on a project which want to include two model, a voice recognition in the first then a image classification. After i trained two model, how can i combine them into a same arduino program. Is it possible to do it?
This is technically possible. However, it requires more surgical work on the C++ side.
One simple solution can be done as follow:
- Download the two libraries into different folders.
- Create a C++ wrapper around the inference functions in each model and compile each library to a specific
.afor static linking
- Call the wrapper functions from each library in your main function.
The more complex solution is to find a nice way to combine the two metadata, and to be sure that the DSP for each model is called correctly. This will require a lot of modification for the