In NanoEdge AI Studio version 5.1.0, the multi library feature was removed. A huge amount of work was done on the back-end, and this feature was not yet ready. We plan to bring this feature back natively in the next major release of NanoEdge AI Studio. In the meantime, we want to provide users with a solution to continue using the multi library.
1. Merge library script
1.1 Description
You can find attached to this article a Python script to merge NanoEdge AI libraries of any kind (anomaly detection, classification, etc.).
This script merges N NEAI static libraries (.a) into a single one, deduplicating identical objects and suffixing global symbols from each additional library’s unique objects to avoid symbol collisions when linking everything into the same project.
Command-line usage example:
python merge_neai_libraries.py --base library1 lib2:library2 lib3:library3 --output libneailibrary1, library2, and library3 are the paths to the folders containing libneai.a (the unzipped output folders generated by NanoEdge AI Studio).
1.2 Use case example
In NanoEdge AI Studio, first compile all the libraries you need in the Deployment step.
Note: you need to unzip the generated files and put each library into its own folder.
In this example, we use 2 libraries:
- An extrapolation library
- A classification library
Then open a terminal and run the following command:
python merge_neai_libraries.py --base libneai_regression/ lib2:libneai_classification/ --output libneai
You should get this output:
Finally, you get a new folder containing a single libneai.a, which is the result of merging all the libraries, and a .h header file for each library containing the functions to use in your embedded project.
