CARE_on_image.py does not currently accept asynchronous execution through MATLAB. The --matlab parameter still exists, but raises exception when used. This parameter is for MATLAB and Python to be running asynchronously and continuously passing data between them. This is an ideal approach because it means that the PythonInferenceEngine--which requires overhead to initialize--could be initialized once and only once.
Previously, the approach was to use the MATLAB workspace as a repository to communicate between MATLAB and Python processes, with variables such as
- PYTHON_INFERENCE_ACTIVE
- PYTHON_SETUP_COMPLETE
- PYTHON_CURRENT_IMAGE
However, this approach failed because MATLAB does not support truly asynchronous modifications to the workspace. Any time a process is running in the MATLAB window, Python cannot make changes to the workspace.
The asynchronous approach could still be possible through a different approach! There are two options for this:
- Use a locally hosted server as a communication vehicle
- Use a temporary
.json file as a communication vehicle
CARE_on_image.pydoes not currently accept asynchronous execution through MATLAB. The--matlabparameter still exists, but raises exception when used. This parameter is for MATLAB and Python to be running asynchronously and continuously passing data between them. This is an ideal approach because it means that the PythonInferenceEngine--which requires overhead to initialize--could be initialized once and only once.Previously, the approach was to use the MATLAB workspace as a repository to communicate between MATLAB and Python processes, with variables such as
However, this approach failed because MATLAB does not support truly asynchronous modifications to the workspace. Any time a process is running in the MATLAB window, Python cannot make changes to the workspace.
The asynchronous approach could still be possible through a different approach! There are two options for this:
.jsonfile as a communication vehicle