Note: Please read the main documentation and also follow the setup instructions before attempting to compile custom models.
The following Jupyter Notebooks shows how to compile a custom model: tutorials/tutorial_classification.ipynb, tutorials/tutorial_detection.ipynb. These Jupyter Notebooks can be invoked by running run_tutorials_pc.sh
The following script can be modified to run custom model compilation from command line: scripts/benchmark_custom.py. This can be invoked by running run_custom_pc.sh.
Runt the script run_package_artifacts_j7.sh to package the artifacts for use in the target device.
In this repository, we have created a symbolic link modelartifacts to the actual modelartifacts folder in edgeai-modelzoo. You may want to rename that symbolic link to something else before attempting to compile custom models - in order to avoid confusion with existing pre-compiled artifacts.
A CSV report containing all your benchmarking results is generated by running scripts/generate_report.py