- main_deployment.py, the main file that needs to be executed to perform the deployment and testing of the models.
- test_models_on_target.py, all the required functions for testing the deployed C code using GVSoC.
- utility.py, some useful functions used throughout the process.
After training and quantization all the required files, stats, and values are saved in the /training_quantization/ folder.
Therefore, during the execution, the /training_quantization/deep_learning_config.ini is read to retrieve useful information, such as quantization values and versions of the models.
Please keep in mind that if you want to perform the deployment on older trained versions of the models, you would have to modify the QUANTIZATION_NAVIGATION and QUANTIZATION_CLASSIFICATION parameters in the file accordingly.
You can find the inputs and output quantization values in the folders on which your models are saved, as a .txt file.
To perform the deployment, execute the following command: python3 main_deployment.py
During the execution, two new directories are created and filled with the respective deployed models in C code.
- classification_model_quant/
- navigation_model_quant/